The Diachronic Self, Chapter 1: Causal Conditions for Continuity

PART 1: REDUCTIONISM, CAUSE, AND IDENTITY

Chapter 1

CAUSAL CONDITIONS FOR CONTINUITY

INTRODUCTION

For the purposes of this chapter I am assuming reductionism to have been established as a component of a theory of personal identity or survival. This chapter will examine differing accounts of causal conditions for personal continuity or identity and argue for one of them.

            The assumption of reductionism throughout the following can summarized as this: My identity, survival, or continuity can be understood as reducible to certain other facts; these are facts about psychological connectedness and continuity.[1] Parfit refers to these facts as the “R-relation.” I am not something ontologically separable from the R-relation. The Reductionist claim is: “A person’s existence just consists in the existence of a brain and body, and the occurrence of a series of interrelated physical and mental events.” (Parfit, 1984, p.211.)

            Even if we agree on Reductionism there are still a number of differing views of personal identity available. I will set out these views in descending order of how restrictive is the causal condition each one specifies. To clarify the differences between the possible views I will provide five thought experiments and will show the differential response of the theories to these cases. Further clarification will come from setting out the different types of cause that can support continuity of the R-relation and examining what assumptions each theory makes about the various types of cause.

            Finally, I will provide several arguments in favor of the Conservative Interpretation of the Widest Reductionist View (CIWR). My conclusion will be that personal identity can survive the absence of normal, reliable, or direct causal connections between one stage of a person and a continuer. Some kind of causal connection between stages of a person is required for them to count as stages of a single individual. This chapter will provide a foundation for later chapters of the dissertation which will examine transformation of self, i.e., various ways in which a self can change over time and the significance of these changes, and how they may be assimilated into the self.

DIACHRONIC IDENTITY

The conception of the self being developed here Armstrong has called a relational view (or what has been called a perdurance view), as opposed to an identity view (or endurance view).[2] We can look at a person who persists over time and divide up their life into any number of non-overlapping temporal stages or phases. An identity view, in Armstrong’s sense, would hold that these phases are identical with one another in some numerical sense. The identity view treats temporal parts differently from spatial parts of a thing: Spatial parts are clearly different parts of a particular, P; they are obviously not identical with one another. (The asymmetry between the identity view’s treatment of spatial and temporal parts provides grounds for an objection to that view, according to Armstrong.) The relational view, by contrast, treats spatial and temporal parts symmetrically. On the relational view, non-overlapping phases of some perduring particular, P, are not identical in any sense. These phases are simply different parts of the same thing. That thing is constituted by those temporal parts and their relations to each other and to other particulars. The account of personal identity or continuity presented here is relational in this sense. The self—the diachronic, continuant self—consists of its temporal stages or phases and the relations between them. The particular relation, in this case, is what Parfit calls the R-relation: Psychological connectedness and continuity.

            At this point it would be sensible to explicitly stipulate how I shall be using the term “self”, before confusion arises. Some people use the term to refer to the temporal phases or person-stages of the continuant person. Others use it to refer to the diachronic particular constituted by its phases and their relations. It is especially important to define my usage since I am building on Parfit’s theory of psychological reductionism, and so might be assumed to be following his usage. Parfit sometimes presents his discussion of personal identity or survival in terms of successive selves.[3] Looked at this way, the continuant person is made up of, or can be regarded as, a series of successive selves. Any two temporally contiguous selves are highly psychologically connected, whereas widely temporally separated selves may be only very weakly connected. However, I will not adopt this usage. I will use “self” to refer to the continuant, perduring, diachronic individual. Its constituent temporal parts I will refer to as person-stages, person-phases, or phases of the self.

Perduring, continuant, diachronic person = SELF

Transient, temporal part of person = PERSON/SELF-STAGE or PHASE

My reason for preferring this usage will become more obvious as this chapter proceeds. Essentially, I believe that the contrary usage—using “self” to refer to the temporal phases—reflects and encourages too heavy a weighting of the significance of these phases, and devalues the importance of the continuant self. This difference in emphasis between my transformationist interpretation of psychological reductionism and Parfit’s version will show up in the sections on the importance of projects and values to the continuity of the person.

            Apart from the unwanted emphasis on the short term resulting from equating self with person-phase, such an equation can easily give the misleading impression that there really are relatively distinct selves. We may talk of the infant self, the child self, the adolescent self, and the adult self, and think of the continuant self as the temporal concatenation of these distinct and successive selves. Nevertheless, this obscures the fact that we rarely find anything resembling a clear line or sudden transition from one such ‘self’ to another. Those four terms are merely loose references to person-phases; the borders they draw can be arbitrarily moved around with some latitude. For instance, we may draw the line marking the change from the adolescent self to the adult self at 13 years (as do Jews and some other cultures), or at 16, or 18, or 21, or the age (whatever it turns out to be) when some specified qualities have been developed.

            Instead of talking in terms of successive selves, I shall stick with the more basic language of degrees or extent of psychological connectedness. If I need to refer to earlier or later instances of a person, I will also use the terms self-stage or self-phase (or person-phase). In other words, I will replace a series of successive selves with a spectrum of connectedness. Connectedness can be measured in two ways giving different answers though, in common with everyone else, I will use the first way. The two ways differ in what to use as the standard of connectedness degree. The first and obvious way is to ask how much of the earlier phase (A) survives or continues on in the later phase (B). (Rather than earlier and later phases, A and B could be original and duplicate selves.) Take the case (illustrated in Figure 3 below) where half of A’s characteristics are shared by B. B, in addition, has a great many characteristics not shared by A. According to the first way of measuring connectedness, A and B are 50% connected (or A is 50% connected to B). Another way to say this is that 50% of A is subsumed in B. The second way measures connectedness in terms of B. We would then describe Figure 3 as a case where connectedness was very low (say 1%) because A has only 1% of B’s characteristics. When A and B represent earlier and later selves (as they will throughout this chapter), only the first way of measuring seems useful. However, If A and B are taken to be an original self and a copy, the second way will be useful, especially when B thinks about the situation. Henceforth, I shall be assuming connectedness is measured the first way, in terms of the earlier self.

            Determining the degree of connectedness will not suffice to tell us all that we need to know about our earlier and later phases if we (earlier phase) are to make sensible decisions about allocating present vs. future costs and benefits. The same degree of connectedness may attach to situations that are not equally desirable. Knowing only that self-phases A and B are 50% connected (for example) leaves out much information about our relation to the later phase. The statement that A and B
DIAGRAM HERE

are 50% psychologically connected could represent any one of three possible propositions (each of which are represented in the diagrams):

            (1) B has 50% of A’s characteristics, but no characteristics that A doesn’t have, i.e. B is a subset of A. (Figure 1)

            (2) B has 50% of A’s characteristics, and 50% of B’s characteristics are not shared by A. (Figure 2)

            (3) B has 50% of A’s characteristics, and only a small fraction of B’s characteristics are shared by A. (Figure 3)

            We can use set diagrams to clarify the ways in which two individuals may be psychologically connected. A and B may stand for the earlier and later person-stages of a continuant individual (and this is the interpretation I will be using). However, A and B could also represent two individuals, each of whom is a survivor of the original. B could be a copy of A—a copy of more or less fidelity, or who has psychologically diverged over time from A.

            In Figure 1 the earlier self-phase, A, possesses all the characteristics of the later phase, B, but B has only 50% of the characteristics of A. In this case, the later self-phase is a degenerate continuer of A. B has learned nothing new, acquired no new memories, formed no new intentions or dispositions, and values only what A valued, yet has lost half of what made A who he was. In Figure 2, the later self-phase retains 50% of A’s characteristics, but also has about as many new characteristics. In Figure 3, the later self-phase retains 50% of A’s characteristics, but these are now an insignificant fraction of B’s total psychological features. This situation might be realized if A is an infant and B an adult self, or if A is any person of today and B a person who, due to advances in gerontology, has lived for many centuries (or their subjective equivalent[4]). B has added many new experiences and memories, and acquired additional dispositions, abilities, and values.

            Each of these three represents a case of 50% connectedness. Nevertheless practically all of us would prefer our future to turn out more like the situation in Figure 2 than in Figure 1, and most of us would prefer Figure 3 to Figure 2. Many accounts of psychological reductionism suggest or imply that it makes most sense to allocate our concern for our future self-phases proportionally to the degree of connectedness. The three cases just described show this to be implausible. The same degree of connectedness may be arrived at in differing ways, and we will prefer some of these to others. The relationship between the metaphysical degree of connectedness and the normative degree of reasonable concern for later stages is thus not a straightforward one. In the later chapter on “A Transformationist Account of Continuity” I will propose several reasons for concerning ourselves with our future phases more than proportionally to the degree of connectedness.

            Having clarified what I mean by connectedness, I will now set out several versions of Psychological Reductionism and explore how they differ in regard to the causal conditions they assume. Without an account of the causal conditions necessary, we will not know when to say that a psychological connection has endured at all.
FOUR THEORIES

The four theories explained here can be thought of as placed along a spectrum of restrictiveness with regard to required causal conditions. The first theory requires the cause of psychological continuity to be a normal one, other theories relax this restriction, and the fourth theory denies the need for any causal connection between a self and its continuer.

            All four Reductionist theories accept the Psychological Criterion. In order to define this, let us review its components. Psychological connectedness is the holding of particular direct psychological connections such as memory links, the connection between intention and action, and enduring dispositions. Psychological continuity is the holding of overlapping chains of strong connectedness.[5] Strong connectedness holds when “the number of connections, over any day, is at least half the number of direct connections that hold, over every day, in the lives of nearly every actual person.” (206)

            The Psychological Criterion: (1) There is psychological continuity if and only if there are overlapping chains of strong connectedness. X today is one and the same person as Y at some past time if and only if (2) X is psychologically continuous with Y, (3) this continuity has the right kind of cause, and (4) there does not exist a different person who is also psychologically continuous with Y. (5) Personal identity over time just consists in the holding of facts like (2) to (4).[6]

THE NARROW REDUCTIONIST VIEW: We might also call this the “Spatiotemporal Continuity view.” The Narrow Reductionist accepts the Psychological Criterion but construes condition (3) narrowly. On this view, the right kind of cause is the normal cause. The normal cause is the continued existence of the brain since our psychological features depend on our brain states. The Narrow Psychological Criterion virtually always coincides with another reductionist view—that of the Physical Criterion. The Physical Criterion says that “a person continues to exist if and only if (a) there continues to exist enough of this person’s brain so that it remains the brain of a living person, and (b) no different person ever has enough of this person’s brain.”[7] The Narrow Psychological Criterion and the Physical Criterion differ in principle in that the former requires, in addition to the two conditions of the Physical Criterion, that there is psychological continuity, that it has its normal cause, and that no one else is psychologically continuous with the person.

THE WIDE REDUCTIONIST VIEW: (Or “Closest Functional Continuer with a Reliable Cause.”) This also accepts the Psychological Criterion but holds that the “right cause” is any reliable cause and not necessarily the normal one. If the memories, intentions, and dispositions that together comprise the R-relation were to be sustained by some reliable process other than the normal activity of the brain, the Wide View would say that identity had been preserved. For example, memory could be preserved on this view by substituting a mechanical replacement for a collection of neurons so long as no change in function occurred. Or, an example from Nozick:

As you are dying, your brain patterns are transferred to another (blank) brain in another body, perhaps one cloned from yours. The patterns in the new brain are produced by some analogue process that simultaneously removes these patterns from the old one… On completion of the transfer, the old body expires.[8]

Here there is no physical continuity of the brain and an abnormal cause is operating to sustain R-relatedness. The Narrow Reductionist View would say the later individual was not the same person as the original, whereas the Wide View says that they are the same.

            Nozick’s Closest Continuer Theory can be seen as a variant on the Wide Reductionist View.[9] The Closest Continuer theory says that something at t2 is the same entity as X at t1 only if it is X’s closest continuer, is a close enough continuer and is enough closer than any other continuer. The only significant difference between the Closest Continuer View and the Wide Reductionist View concerns Parfit’s fourth condition for the Psychological Criterion. This said that “there does not exist a separate person who is psychologically continuous with Y.” This is included to rule out cases which violate the condition for transitivity of identity. If person X at t1 is R-related to both Y and Z at some later time t2, then, according to Parfit’s account of the Psychological Criterion, X cannot be identical to either Y or Z even if, at a still later time t3 there is only one person R-related to X.

            The Closest Continuer theory, by contrast, might allow Z to be identical to X despite there having been a period in which both Y and Z were continuers of X. Nozick’s example is as follows:

Half of an ill person’s brain is removed and transplanted into another body, but the original body plus half-brain does not expire when this is being done; it lingers on for one hour, or two days, or two weeks. Had this died immediately, the original person would survive in the new body, via the transplanted half-brain which carries with it psychological similarity and continuity. However, in the intervening hour or days or weeks, the old body lives on, perhaps unconscious or perhaps in full consciousness, alongside the newly implanted body.[10]

If the old body had died simultaneously with the transplantation, the new body plus half-brain would be the closest continuer. However, so long as the original body plus half-brain remains alive it is the closest continuer. After it dies, the new body plus half-brain becomes the closest continuer. As Nozick asks, “Can its [the old body] lingering on during the smallest overlapping time interval, when the lingerer is the closest continuer, mean the end of the person, while if there was no such lingerer, no temporal overlap, the person would live on. It seems so unfair for a person to be doomed by an echo of his former self.”

            Nozick then offers several possibilities, one of which is that the person moves from the original body to the new body not when the transplant occurs but when the old body dies. (The reasonableness of this view is supported by Nozick’s previous examination and rejection of condition that the identity of Y at t2 with X at t1 depends only on the properties and relations of X and Y and not also on whether there exists a Z which more closely continues X.)

            Nozick does not conclude that this is the right answer; rather he claims that a certain indeterminacy is inherent in identity ascriptions. Our choice in this case depends on whether we structure the concept of identity locally or globally. We needn’t be concerned here with these complications. We can simply note that a Closest Continuer view might allow identity to continue despite some temporal overlap during which there are two close continuers. Of course, if the old body lingered for several years before expiring we would not say that the new body plus half-brain was a close enough continuer (since it would have been going its own way for too long). There seems to be no clear way of saying how long of an overlap is just too much for the new body plus half-brain to count as identical with X.

            If the Closest Continuer theory is taken to allow some degree of temporal overlap then this will conflict with Parfit’s fourth condition for the Psychological Criterion which disallowed any separate person to also be psychologically continuous with the original person. With this one difference however, the Closest Continuer theory and the Wide Reductionist view can be made compatible. If both theories accept a psychological criterion then, with the possible exception of the overlap case, they will come to the same.

THE CONSERVATIVE INTERPRETATION OF THE WIDEST REDUCTIONIST VIEW (CIWR): (Or: “Closest Functional Continuer with an Indirect or Unreliable Cause.”) The Widest Reductionist View interprets the “right cause” third clause of the Psychological Criterion as allowing any cause. So long as the later continuer is caused to occur in some way, and the earlier stage of the person plays a crucial role, even if indirect and unreliable, then the continuer counts as the same person as the earlier person. An example will be given below to illustrate what this might mean.

THE EXTREME INTERPRETATION OF THE WIDEST REDUCTIONIST VIEW (EIWR): (Or: “Closest Functional Continuer with No Cause.”) This actually subsumes two theories, the distinction between which will be illustrated by the fourth and fifth cases below. The two possibilities depend on what “No cause” is taken to mean. What they both have in common is that they allow identity to continue even though the earlier stage plays no causal role at all in the creation of the qualitatively identical later entity.

            “No cause” might mean (a) as just stated, that the earlier entity in no way causes its qualitatively identical successor; however, the successor is caused to exist by something, albeit something unrelated to the earlier entity; (b) the most radical interpretation of the Extreme Interpretation of the Widest View holds that Y can be identical to X so long as Y is qualitatively identical to X, even though (i) X played no causal role in bringing about Y; and (ii) nothing caused Y to come into existence, i.e., Y came about purely randomly. Though these two variants of the Extreme Interpretation differ in principle, they are extremely unlikely to ever differ in practice, and just about everyone will have the same intuitive responses to both. I include both variants for the sake of completeness.

CASES

The following five cases or thought experiments are intended to illustrate the differences between the theories and between different views of the role of causation in the criteria of personal identity. I will refer to them throughout the remainder of the paper, sometimes by their abbreviations: 1:MBT; 2:TT; 3:AHR 4:OPR; 5:LPU.

1: My Brain Transplant: My life is threatened with an as-yet incurable disease. My doctor advises me to have a brain transplant. Cell samples are taken and a decerebrated clone grown at an accelerated pace. My brain is transplanted to the new body and hooked up to it. I am soon walking around in a healthy new body, free of the disease.

2: The Transporter: I step into a We-Beam-U-Quick booth in Los Angeles. My body is scanned and the position and velocity of all my constituent atoms and molecules determined and recorded. The material of my body is atomized—far too rapidly for me to feel anything—and the information specifying the structure of my body is transmitted to Tokyo. A similar booth in Tokyo receives this information and constructs a body exactly like the old one, except that new atoms (but of the same elements) are used. I walk out of the booth happy to have made it so quickly to Japan, avoiding the crush at the airports still used by conservative folk afraid of temporary disintegration.

3: A Heroic Reconstruction: In 2330AD a group calling itself the Order of Universal Immortalism (OUI) declares that its centuries old ambition is finally realizable. Technology has rendered death a matter of choice. Yet, they claim, the death of people in past centuries is still a tragedy that ought to be remedied. Fortunately Personality Tracking and Reconstruction Technology has matured to the point where it is now possible to reconstruct their personalities. Into a Super-Reconstructor Computer is fed information gleaned from the brains of people who knew the unfortunate persons, in addition to the results of extremely complex tracing of causal connections back in time. The 22nd century overthrow of quantum mechanics with its uncertainty principle and replacement by Quantum Super-Determinacy, has allowed (with the aid of vast computational power) the precise determination of the state of the universe at all past times. OUI scientists recently successfully reconstructed the information uniquely specifying the personality of a previously deceased person and embodied that information in a living body. The reconstructed person, Francis Bacon II, expressed surprise at finding himself in the 24th century, but was pleased to see that his scientific method had been able to accomplish so much.

4: Omega Point Resurrection: As proposed by Frank Tipler,[11] at some vastly distant time in the future, our distant descendents, now in control of all matter and energy in the universe, set about resurrecting the dead. To ensure reconstruction of everyone who ever lived, despite the loss of traces of most of them, they create duplicates of every possible person. One of these duplicates is exactly like Francis Bacon, all the way down to his quantum states, as he was shortly before he died.

5: The Luckiest Person in the Universe: From our Los Angeles correspondent: They say that L.A. is the city where dreams come true. But even hardened Angeleno’s were stunned by what happened today in front of City Hall during a press conference. In plain view of a crowd of people, and recorded by numerous cameras, a man identifying himself as Francis Bacon, materialized out of nothing. Investigators have ruled out any trickery and have been unable to explain this man’s appearance. “Bacon” has an appearance in keeping with historical records and has offered details of the historical character’s life and many detailed historical facts which have impressed historians specializing in the area. One physicist has suggested that, although such an event is extremely unlikely to occur even once in billions of years, it is possible that Mr Bacon is the outcome of a completely random and

acausal quantum process. Atoms in the area might have just happened to have simultaneously randomly changed state in such a way as to produce this mysterious man.

CASES AND THEORIES

The four theories outlined above can be distinguished more clearly by showing how they handle these five cases. The Narrow Reductionist View will allow that there is continuity of personal identity in My Brain Transplant, but in no other case. Continuity of the brain is the normal cause of psychological continuity and a brain transplant preserves the brain. In the other four cases, the brain—as a physically continuous entity—does not endure.

            The Wide Reductionist View would agree that identity survives the brain transplant, but would go further in making the same judgement in The Transporter case. In this second case, the R‑relation is preserved by an abnormal though reliable cause. The brain structures embodying personality are transferred to a newly constructed brain by means of an automatic process that reliably preserves the original patterns. (If this were to become common practice, would the Narrow View come to regard it as a normal means of continuity? If so, does that mean that transporters, when uncommon, fail to preserve identity but do preserve it once they are used regularly?) The Closest Continuer version of the Wide View might allow identity to have continued even if the old brain had not been disassembled immediately upon creation of the new brain at the remote location.

            The Wide Reductionist would describe the case of A Heroic Reconstruction as one where Francis Bacon ceased existing, and later a different person just like him appeared. The Wide Reductionist requires a reliable cause to operate in maintaining R‑relatedness. As discussed below, this third case involves decisions being made by other people such that the continuation of R‑relatedness is precarious and is not governed by a reliable process. The fourth and fifth cases go further in the direction of unreliability and so would not be accepted as instances of the persistence of personal identity by the Wide Reductionist.

            The Conservative Interpretation of the Widest Reductionist View (CIWR) would hold that personal identity had continued in the first three cases. Since this theory requires for personal identity only that there be some causal relation between the earlier stage and the later one, whether normal, reliable, or unreliable, Francis Bacon has continued on as the same person in A Heroic Reconstruction. The new Bacon is qualitatively identical to the historical Bacon, and the later Bacon’s existence is dependent on the existence and qualities of the earlier Bacon. The Universal Immortalists recreated Bacon because of his previous existence, and they made him the way they did because of the way he was. CIWR would describe the situation as one in which Francis Bacon jumped through time from the 17th Century to the 24th, in principle just as if he had been in a coma for seven centuries.

            Despite the apparent looseness of its causal condition for identity, CIWR would reject the last two cases as instances of continuation of the same person. Although qualitative identity is maintained in Omega Point Resurrection and The Luckiest Man in the Universe the earlier Bacon plays no role either in the coming into existence or the qualitative identity of the second Bacon.

            The Extreme Interpretation(s) of the Widest Reductionist View (EIWR) counts the first four cases as instances of the same person continuing. They require only that the later Bacon-like individual be qualitatively identical (or sufficiently similar) to the earlier. No causal connection between the earlier and later Bacons is necessary. I should note here that if Bacon II is qualitatively the same as the original Bacon just before he died, there will be some overlap. On the Closest Continuer variant of the Extreme Interpretation this breakdown in transitivity does not disrupt identity. According to Parfit’s version in which identity requires that “there does not exist a different person who is also psychologically continuous with Y”, the matter is open to interpretation. Since in the example as described, there is no actual temporal overlap, Parfit’s condition has been fulfilled and identity is maintained. However the spirit of Parfit’s fourth, no-overlap, condition may disallow even the non-temporal overlap. If so, then we could redescribe the cases such that Bacon II is created in the same state that the original Bacon had just died in, and then returned to consciousness. Alternatively, we can throw out the no-overlap condition and regard the Extreme Interpretation as an account of personal continuity (Parfitian survival) rather than identity.

            The Moderate interpretation of EIWR (M-EIWR) differs from the Radical interpretation of EIWR (R-EIWR) over the fifth case. M-EIWR requires that there be a cause of the later Bacon, though it need not be a cause relating the earlier to the later person. R-EIWR rejects even this causal condition, requiring only qualitative continuity. R-EIWR holds that the randomly created Bacon in The Luckiest Man in the Universe is the same person as the historical Bacon. I will argue below that there is no important difference between M-EIWR and R-EIWR in practice, and little theoretical difference.

CAUSAL CONDITIONS

All the theories examined here are reductionist. Where they disagree is over the type or necessity of causation in personal identity. In order to decide between the theories it will therefore be helpful to more precisely distinguish the different types of cause at issue. After setting these out I will give five arguments in favor of abandoning all but the weakest causal condition. This will support a view as radical as the Conservative Interpretation of the Widest View (which requires an indirect cause of any kind), but not the Extreme Interpretation.

            Causes can set out along three spectra: (1) From normal to any reliable to any to no cause. (2) From internal to external. (3) According to degree of directness: From direct to indirect to independent cause.

            (1) (a) Normal Cause: The requirement of a normal cause is embodied in the Narrow Reductionist View. The Physical Criterion would take this to mean the standard continuity of a human brain. The Psychological Criterion would add other, more complicated, conditions such as a coherent development of later psychological traits out of earlier traits. Continuity of the brain, considered on a gross level, is an insufficient specification of a normal cause on the Psychological Criterion since the brain may continue but its states might be radically disrupted in a way that introduces a sudden and permanent change in personality. This might come about as a result of some neurochemical shock, or due to an intense psychological shock.

            Clearly the Narrow Reductionist has to make choices about what is to count as a normal cause of psychological continuity. Some might count a sudden and profound psychological change caused by the interaction of intensive “brain-washing” with the existing personality as sufficiently abnormal to destroy personal identity while others might not.

            Another question to be answered by the Narrow theorist is whether or not “normality” is to be temporally indexed: Does “normal” mean “normal for 1990”, “normal up to 1990”, or “normal considering past, present, and future”? The problem is that what is normal now may not always be so. A few hundred years ago, the normal cause of good vision was having good natural eyes and no good vision existed where eyes were faulty. Now, however, good vision is caused almost as often by artificial devices and processes such as glasses, contact lenses, radial keratotomy (and soon corneal sculpting). Perhaps one day unaltered brains will not be the normal cause of psychological continuity. Just as we now often replace other faulty organs with artificial organs, we might eventually replace parts of the brain with functionally equivalent artificial parts (biological or mechanical), and perhaps even the entire brain might be replaced (gradually over time or all at once) with an artificial brain. If this were to occur would the Narrow Reductionist say that the new method was an abnormal cause and didn’t continue identity, or would she say that it is what is currently normal that counts? If the latter, then why should we talk of a normal cause at all? Why not require simply any reliable cause? Such indeterminacy in the normality condition strongly suggests that this view is parochial and inessential.

            (b) Any Reliable Cause: This is the condition minimally required by the Wide Reductionist View (and Nozick’s version of the Closest Continuer Theory). A reliable but abnormal cause is illustrated by The Transporter (if we assume either that transporters are unusual or that “normal” refers to historically normal). In the transporter case psychological continuity is reliably maintained since the process is (by hypothesis) both automatic and about as reliable as the brain in ensuring continuity.

            To be fanciful: Suppose that we are mistaken in our materialistic view that our brains’ operations are determined by physical law and that in fact the appearance of causal connectedness has always been brought about by “God.” (That is, suppose God sustains the universe in existence rather than merely having created it with the property of self-sustenance.) The normal and reliable cause of psychological continuity is then God’s intentions and power (assuming that God has a long-term intention to maintain standard operations). One day God decides that he is bored with existing and decides to cease existing. However, he feels responsible to his creation (or emanation) and so changes the nature of the universe so that it is now self-sustaining just as we materialists[12] always thought it was. The new cause of our psychological continuity is not (yet?) the normal cause but it is just as reliable as the old cause. The Wide Reductionist view sees this case as unproblematically one where the reliable but non-normal cause is sufficient for continuity of personal identity.

            (c) Any Cause: This is illustrated by A Heroic Reconstruction and is the minimal causal requirement of the Conservative Interpretation of the Widest View (CIWR). More accurately this condition says: “Any relevant and sufficient cause connecting the earlier entity with the later qualitatively identical entity.” This condition is less demanding than the normal or any reliable causal conditions since the relevant cause need be neither normal nor reliable. The dropping of these two conditions is shown by A Heroic Reconstruction. Here the new Bacon is clearly produced by a very abnormal cause. The causal connection between the old and new Bacons is unreliable since it is mediated by the intentions of other agents. If the Universal Immortalists had held different beliefs and/or desires then the new Bacon would not have come into existence.

            Though the requirements of normality and reliability are dropped on the “any cause” view, there remains the requirement that there be a causal connection between the earlier person and the later one. Nor can this be just any causal connection. In the case of Omega Point Resurrection, imagine that the second Bacon happened to pick up some old writing of the earlier Bacon. The later Bacon then carries out an unfulfilled intention expressed in that writing. This kind of connection would not be sufficient for continuity of personal identity on the “any cause” view. In this example, though there is a causal connection between the earlier and later Bacons, it has nothing to do with the existence or qualitative identity of the later Bacon (who was produced without any knowledge of the original). The “any cause” condition requires that the causal connection be one necessary to explain the coming into existence of the qualitatively identical later person and sufficient to explain the qualitative similarity of the later person.

            (d) No Cause: This drops even the requirement that the earlier entity be causally connected in any way to the later one. This corresponds to the Extreme Interpretation of the Widest View. There is a moderate and radical version of the “no cause” view corresponding to the Moderate and Radical interpretations of the Extreme Interpretation of the Widest View (M-EIWR and R-EIWR). The radical interpretation takes “no cause” to mean not only that there is no relevant cause connecting the earlier and later qualitatively identical persons but also that there need be no cause of the later person at all, i.e., the person could have come into existence without necessitating prior causes.

            (2) Internal to External: An internal cause of personal continuity is a cause operating within the physical system that instantiates the person. The continuation of psychological relatedness over time when supported by the continued internal operations of the brain is a case of an internal cause. Internal causes are likely to be normal causes, but there could be abnormal internal causes such as the gradual replacement of brain tissue by artificial components.

            An external cause exists where the continuity of personality results from something other than the internal workings of the physical support system, such as in The Transporter. If the continuation of my personality depended on continual acts of God (as in some interpretations of Berkeley’s views) then it would have an external cause.

            The point of distinguishing a reliability spectrum from an internality-externality spectrum is that, for many people considering the cases, an internal unreliable cause is much less disturbing than an external unreliable cause. Suppose a new disease became epidemic; in most persons this disease leads to rapid deterioration of neural function, thereby destroying personality. Although the brain in such a situation would not be a reliable cause of psychological continuity, few would feel that the identity of those not succumbing to the disease was threatened. By contrast, the case of A Heroic Reconstruction, which could be no less reliable a cause of continuity than exists in the diseased-brain society, will lead far more people to doubt the continued identity of persons. Doubts about continuity of identity increase independently with both increasing unreliability and increasing externality of causes.

            (3) The Directness of a Cause: This is likely to be related to its reliability. Directness specifies how many steps are necessary to the process that continues the qualitative identity. The more steps there are ceteris paribus the less reliable the cause of continuity will be. However, other factors may not be equal. If the technology and construction of transporters were excellent, then the indirectness of continuation by transporter may be no less reliable than continuity resulting from normal brain operation.

            A direct cause need not be normal: So long as the earlier state of the physical system supporting personality directly leads to the next state of the system it doesn’t matter whether the system is the normal one. A partially computerized brain might just as directly cause later states of the system.

            An indirect cause (as in The Transporter and A Heroic Reconstruction) may or may not involve actions on the part of other persons. If actions and decisions of others are required in order to maintain continuity then the cause will probably be more indirect and less reliable. The cause of continuity will become increasingly indirect as more agents must decide whether to participate in maintaining a person’s continuity. Every time continuity depends on someone’s decision, the unreliability of the continuity increases. An independent cause is the same thing as the moderate interpretation of “no cause.” In this case the continuer of the earlier person has a cause that is entirely unrelated to the earlier person.

            The foregoing suggests that the directness spectrum is not as deep as the reliability or internality-externality spectra. Degrees of directness can be reduced to combinations of degrees of reliability and internality/externality (especially the former) whereas the latter two spectra cannot usefully be reduced to each other.

WHICH CAUSAL CONDITION?

The foregoing has consisted of conceptual clarification and differentiation. I want to distinguish the various views as sharply as possible in order to see precisely what is at issue between them. Almost everyone (every reductionist at least) will be satisfied that personal identity is secured so long as the normal continuity of the Narrow View is maintained. But beyond that point intuitions differ over where the cut off point lies. Some will reject The Transporter case though most reductionists are willing to accept that as preserving personal identity. Quite a few will draw a line between The Transporter, with its unusual but reliable and fairly direct cause on the one hand, and all the later cases on the other. Others will feel that the first three cases go together since there is some important causal connection between the earlier and later persons. Very few people will react differently in comparing Omega Point Resurrection and The Luckiest Man in the Universe. I will provide several arguments that together I take to strongly support the supposition that personal identity exists in at least the cases of My Brain Transplant, The Transporter, and A Heroic Reconstruction. Before arguing that we should be willing to go as far as AHR and the Conservative Interpretation of the Widest Reductionist View, I will argue that we should not abandon all causal conditions. That is, we should reject the Extreme Interpretation of the Widest Reductionist View. Omega Point Resurrection and The Luckiest Man in the Universe are cases in which personal identity is not maintained.

Is a Causal Condition Necessary?

According to the Widest Reductionist View any causal connection that preserves connectedness is sufficient for the very same individual to continue. I distinguished two variants of this view—the Conservative and Extreme Interpretations. The Extreme Interpretation allows “any cause” to include no cause. That is, psychological connectedness may persist over a causal discontinuity. This section will argue against the Extreme Interpretation of the Widest Reductionist View.

            My view, like Parfit’s, is that it is the effect rather than the kind of cause that matters in these cases. Thus radically different mechanisms like brain transplants, teletransportation, or personality reconstruction (as in A Heroic Reconstruction) can maintain the connectedness necessary to personal continuity and identity. As I will show (mostly in the final chapter), some kinds of changes and causes of changes are more compatible than others with a judgment that the same person continues. If we meet a friend whom we have not heard from in ten years, and we are startled by the enormous difference in their personality, it will matter to us how they became who they are now. If they developed into their current self in normal ways we will not hesitate to believe they are numerically identical with the person we used to know (or: this stage is a continuer of the person-stage we used to know). If we discover that the personality of the friend we once knew was destroyed and replaced with an implanted persona (as portrayed in the movie Total Recall and Phillip Dick’s original story “We Can Remember It For You Wholesale”), we may not regard the former person as having survived.

            Although I take a liberal view of the kinds of underlying causes necessary for the same person to continue to exist, I must challenge Daniel Kolak and Raymond Martin’s no-cause view in “Personal Identity and Causality: Becoming Unglued.”[13] Kolak and Martin present a range of cases intended to make the reader “walk across the bridge” from normal, reliable causes to no cause. Before developing my positive account of causal conditions for identity, I will critically examine Kolak and Martin’s examples, showing why I think that a causal requirement should be retained. In order to have a clear foil against which to argue, I will first need to ensure that their examples genuinely are cases of the absence of causal connection.

            The two examples purport to describe a situation in which sequences of events follow one another just as they do in our universe (as we understand it), but where there is no causal connection between events. The first example offered by Kolak and Martin asks us to assume that science one day discovers that the universe is causally just like the situation in their MOC Match Example. This involves supposing that

the entire universe as we now know it flickers on and off every instant. Between instants, some sort of cosmic randomizing device creates an infinite number of sequences and then selects and conjoins with the others only those which correspond in a normal way to the previous instant. In that case, what exists at any one instant did not causally arise out of what existed at the previous instant. It is not likely, to say the least, that science will discover this. But if it did, would it make any difference to our concept of personal identity, or would we go on just as before? We believe that we would go on just as before, and that the fact that we would casts doubt on the causal condition. (p.344)

            This example fails to support the claim that a causal connection of some kind is unnecessary. One problem we face immediately is in understanding the workings of the universe as their example describes it. I can see two significantly different ways of interpreting this example. According to the first, the universe described has two main components: A cosmic randomizing device (CRD) that seems to be the active component, and the universe as we perceive it, a universe that really consists of consecutive universe-slices created by the CRD. The CRD is quite distinct from the universe-slices, being itself stable and not subject to the constant flickering into and out of existence. If the CRD is to generate universe-slices in a normal way, as posited, then it must contain an internal model at least as complex as the universe. Without a model this complex the CRD would be unable to “correctly” select the configuration that matched the previous universe-slice and bring it into existence. In this situation it appears that the universe as we perceive it—the sequence of universe-slices—is irrelevant to the progression of events; everything of importance takes place within the CRD where the model of the universe evolves smoothly. The external universe-slices appear to be nothing more than epiphenomena; as such Kolak and Martin’s first example becomes no different from their second, making my criticisms of that example applicable to this.

            Leaving this interpretation aside, the example admits of another interpretation. The most natural interpretation, and the one that accords with Kolak and Martin’s description of the MOC Match Example from which it is adapted, is one where a causal connection is maintained. Though one universe-slice does not directly cause the next one to match its configuration, each subsequent slice is configured by the CRD because that configuration matches the foregoing slice. If it weren’t for the earlier universe-slice and its unique configuration, the CRD wouldn’t have generated the later slice. If the CRD is causing successive states of the universe to match up then the example does not illustrate a situation where we believe persons to have persisted in the absence of causal connections.

            For the example to meet the no-cause requirement we must reformulate it. The best way is to exclude the deus ex machina, the CRD, leaving the example otherwise unaltered. Given this third interpretation, the example is genuinely one where there are no causal connections between (qualitatively non-identical) states of the universe.

            Their second case asks us to

Suppose that what we ordinarily take to be causally related stages of a continuing person are not related as causes and effects of each other but are rather epiphenomena of some underlying process…

In our example it is our bodies-plus-our-conscious-lives, the entire complexes of what most of us think of as ourselves, that are causally impotent. This could happen if there were very small temporal gaps between adjacent, “instantaneous” person-stages, and no causal influence were transmitted across these gaps. Person-stages then would merely appear to be causally related because the underlying process insures [sic] that they give that appearance. Further, suppose that the underlying process is not such that if we understood it, we would want to include any of it into what we regard as a person.

The description of this example is brief, leaving it open to interpretation. However, only one interpretation is strictly consistent with the authors’ intent to remove all causal connection between person-stages. This interpretation says:

During the fleeting span of each instantaneous person-stage, no change occurs and no causes exist. The underlying process removes each stage from existence and replaces it with a slightly different stage, producing the appearance of change and causal connection between stages.

This makes the underlying process the Prime Mover and the clearly makes the stages epiphenomena. There are causes operating, even though they are radically different from those allowed by our current physical and biological models. In these cases, between the stages of what we now think of as distinct, complete, and continuing persons, there is no direct causal connection. If the epiphenomenalist example turned out to be actual, Kolak and Martin claim that such a discovery would not affect our everyday lives; “we would continue to associate person-stages with persons in the same way that we do now.” (345) I accept this conclusion; I would continue to believe in enduring persons. Nevertheless, I deny that this example should lead us to reject causal conditions. Not only does the example fail to justify throwing out all causal conditions, it does not provide a reason to move beyond even the Narrow Reductionist’s conservative position. Instead of rejecting causal conditions, we would have to reformulate our conception of selves.

            If we discovered such an underlying process, and that process was recognizable as a person, then we would merely have to accept that we are located differently than we had believed, just as someone studying neuroscience might realize that consciousness resides in the brain and not the heart. We could explain the illusion of being centered in our heads as being due to the location of most of our senses, just as today’s experimenters strapped into virtual reality gear have had the experience of viewing their bodies from the outside, and feel that they are located at their point of view. Relocation of the self is easy enough to imagine, but Kolak and Martin explicitly rule it out. They stipulate that the underlying process is not itself a person. This leaves us with the option of redrawing the boundaries of persons. Rather than being located only in brains and bodies, as we had thought, we would see ourselves as spread out between brains, bodies, and the underlying process, wherever it was located. Perhaps the newly discovered process is a field composed of a novel form of energy, spread out over miles, overlapping with the fields of other persons. If so, we would come to see ourselves, contrary to our untutored perceptions, as much larger, and as being able to share space with another person in a way that our bodies cannot. The continuing experiences of our reconceived selves would be caused not by the brain, or not by the brain alone, but the underlying-process-brain combination. Whereas today we believe one thought (or emotion, or intention) to follow from another because of causes found in the brain, after the paradigm-shattering discovery we would believe that thoughts follow one another because of causes embedded in the underlying process. The causes were there all along, even though we were in ignorance of them, therefore they are the normal and (presumably) reliable causes. The discovery, startling as it would be, would give us no reason to move even beyond the Narrow Reductionist View.

            The foregoing might not satisfy Kolak and Martin. They stipulate in their epiphenomenalist example not only that the underlying process is not a person, but the much more restrictive stipulation that “the underlying process is not such that if we understood it, we would want to include any of it into what we regard as a person.”  The underlying process is not a person, nor is the process plus apparent (causally unconnected) person-stages. This move is intended to force us to the view that, since our experience tells us that we are continuing persons, persons must be able to persist despite an absence of causal connection between person-stages. Syllogistically, the argument boils down to this:

(1) We might discover that person-stages are epiphenomena of an underlying (non-personlike) process and lack any causal relation to one another.

(2) In that situation we would still believe persons to exist.

Therefore, causal connection between person-stages is not necessary for continuation of persons.

            That is, we exist, possess consciousness, and our experiences proceed in an orderly fashion, yet each thought, each action, each feeling, each person-stage, has no causal relation to the one before. If I accept Kolak and Martin’s analytical procedure, I must reject one of the premises since I maintain the contrary of the conclusion. Although I have reservations about the procedure I will not challenge it because that would require arguments about the proper use of thought experiments, and that would take me too far afield.  Even accepting their procedure, I do not have to accept the conclusion of the argument. If we were to uncover the truth of this epiphenomenalist scenario, I would conclude that there were no persons.

            Our notion of persons involves enduring entities with certain properties such as (a capacity for) rationality, responsibility, the ability to make choices, foresight, and (a capacity for) self-restraint. None of these properties could exist in the epiphenomenalist scenario. If I form an intention to do X at one moment, it can have no effect whatsoever on whether I (or the appearance of a continuing “I”) later do X. The succession of time-slices is determined by the non-personal underlying process so that intentions will be followed by later actions that accord with those intentions, yet there is no connection at all between the two. If I (the momentary time-slice) had somehow formed a different intention, the underlying process would nevertheless produce later time-slices according to its program. Any decision I make at any moment can have no effect on “my” future actions. Similarly, even if I had different experiences than I actually have, my memories would be the same—as determined by the underlying process. If I were to consider evidence different from that which I have, I would still reach the same conclusion—the conclusion programmed by the underlying process. Nothing I do now can make any difference to what I will do later. Nothing I have done in the past can have any effect on what I choose to do now. The appearance of a connection between actions and characteristics over time is only an appearance. The epiphenomenalist scenario leaves no room for genuine choices, responsibility, or reasoning. Without these characteristics the mere appearances left over do not remotely satisfy our requirements for personhood.

            Our experience seems to assure us that there are persons and that this belief must persist after the epiphenomenalist discovery. But our belief that we are persons, in the sense standard for the term, is no more immune to revision and rejection than beliefs about the rest of the world. Indeed, in the actual world, a substantial body of research suggests we are systematically deluded about the nature of our experience: see studies of blindsight, and experimentally-backed theories asserting that we are never conscious of the present.[14]

            I conclude that this imaginary example should not lead us to abandon the causal condition. Kolak and Martin are mistaken in their surprising assertion that abandoning the causal condition in the face of this hypothetical discovery is a conservative suggestion. If our successive conscious states were causally unrelated, it is not true that everything else would be left intact. Nonsense would be made of our notions of persons as agents, as responsible in any way for our actions and thoughts, and of personal projects and goals of having any value. Our intentions and memories and other characteristics would have nothing to do with what we did. If we are satisfied that persons only persist in so far as their phases are causally connected, we can move on to consider how liberal we can be in allowing non-standard causal connections to count as maintaining a person’s identity.

ARGUMENT FROM PERCEPTION AND INTUITION

This argument supports a move at least up to the Wide Reductionist View, and has been used by Nozick in defending the Closest Continuer variant of Wide Reductionism. The essential idea is that we perceive the continuation of objects according to a certain schema which is well described by the closest continuer theory. This schema explains our intuitions regarding the diachronic identity of changing objects. The unstated assumption of Nozick’s point is that our conceptual understanding is strongly informed by our perception. If our perception works according to a certain schema then this provides some support to the belief that the schema well describes objects at the conceptual level.

            Nozick describes experiments by psychologist Shimon Ullman intended to determine how and when people classify discontinuous appearances as two appearances of the same object and when as appearances of two different objects. An object was shown moving in a straight line towards a screen and then disappearing behind it. After a short time another object came out of the other end of the screen. The angle of exit was varied, as well the object’s color, shape, and velocity. The results of the experiments fitted the closest continuer hypothesis. For example, if the object exited at a different angle, or after a delay, people assumed that it had been struck behind the screen and was the same object. But if two objects exited the screen, the one with the angle, velocity, etc., of the entering object was thought to be the original object.

            In these judgements of continuity of the same object, it does not matter whether the object is perceived continuously. Nozick thinks that these experiments support both parts of the Closest Continuer theory: (a) That a later object Y is the same as an earlier object X only if Y’s properties grow out of, are causally dependent on X’s properties at the earlier time; (b) there is no other Z at the later time that stands in a closer (or as close) relationship to X than does Y.[15]

            If the argument from perception has weight it supports Wide Reductionism because it supports the view that something is the same thing so long as it is similar and continues sufficiently smoothly from an earlier phase of an object. It seems not to matter that the qualities of the object are not spatio-temporally continuous. However, as Nozick presents it, the perception argument rules out the Extreme Interpretation of the Widest View since this rejects (a). This argument gives some support to the Conservative Interpretation of the Widest View. CIWR weakens the causal condition considerably by allowing quite indirect causes to maintain identity. Certainly CIWR is compatible with the experimental results: The subjects do not know what is happening behind the screen; perhaps some unusual causal process occurs, but one that results in an exiting object with a similar enough trajectory, velocity, or appearance to convince the viewers of its continuity with the object that entered. Although compatible with CIWR, these results do not provide convincing grounds for moving from the Wide to the Widest Reductionist view. The experiment does not test how the subjects’ responses might be effected if they were to consider the possibility that something highly unusual had happened behind the screen. Perhaps the entering object was instantaneously scanned then disintegrated, and an exactly similar object created according to the information gleaned in the scan and put back on the original object’s trajectory. Perhaps the experiments weakly support CIWR in particular in that unusual causes like this could have been operating, yet the subjects still made the judgements they did. However, the experiment was not designed to specifically test reactions to such situations.

            This argument from perception is suggestive and may give some weight to the Wide and even the Widest Reductionist Views, but it is not strong enough to give these views sufficient support by itself. We might doubt that the operation of our normal perception and conceptualization is a good guide to highly abstract problems of identity as in personal identity. This is because of the Quinean and Hansonian point that percepts must be conceptually interpreted, and interpretations may be based on parochial experience and incorrect premises. This first argument is therefore only a prelude to other more powerful arguments.

NEUROSURGEONS AND THE SLIPPERY SLOPE

Personal identity is not defined by an entity’s continued possession of the same collection of atoms. All versions of the psychological criterion of reductionism accept this, as does Nagel’s physical same-brain criterion. The same-brain criterion requires only that enough of one brain persists along a unitary spatio-temporal path. The irrelevance of material persistence to personal identity or continuity is supported by functionalism. Functionalism identifies beliefs, desires, and other psychological states by their causal roles in an economy of such states. My belief-that-p is distinguished by its relation to input, to other psychological states, and to behavior.

            Functionalism supports psychological reductionism (Relation R) over physical reductionist views like Nagel’s.[16] Nagel’s view is that I am essentially whatever is normally the cause of psychological continuity, and this is a brain composed of the usual proteins. Nagel sometimes seems to want to say that it is the brain in certain states, rather than simply the whole brain, that is essential to identity. But even that version of the same-brain criterion conflicts with functionalism. According to functionalists, a belief-that-p might, in principle, be instantiated in a human brain in one way, and in an animal’s, extraterrestrial’s, or computer’s brain in different material. Even in a single, continuous brain, certain neurons might be replaced over time with functionally equivalent elements made of an entirely different material. Even if the majority of the brain were eventually replaced in this manner, functionalism says that the same beliefs, desires, intentions and memories would persist. Since, according to  psychological reductionism, a person just is psychological connectedness and continuity, functionalism supports some version of psychological reductionism.

            Functionalism’s type-token distinction cannot help us decide which variety of psychological reductionism is correct because types at one level of abstraction can be seen as tokens at a different level of abstraction. This means that we cannot be satisfied with simply identifying the R-relation with the functional level rather than the material level. Doing this will not distinguish any of the theories beyond Wide Reductionism from one another since they can all claim to identify a self with a functional level of description.

            This relativity of types and tokens can be illustrated as follows: A belief-that-p is now tokened in me by group-of-atoms A. This group of atoms, at a particular point in time, constitutes neuron group M. Here M, as constituted by specific atoms, is a token. Over time the atoms constituting neuron group M change so that M can be regarded as a type. Suppose that later my belief-that-p is embodied a physically distinct neural circuit N in the same brain;  at a higher level of abstraction the token is now neuron group N and the type is the activation vector space[17] embodied in neuron group N. The activation vector space might, at a later time, be instantiated in an artificial brain with identical function to the biological organ, and where the vector space has identical relations to other vector spaces in the artificial brain as it had when instantiated in the original brain.

            Functionalism supports a move as far as Wide Reductionism but cannot decide between CIWR, and EIWR. Functionalism gives conditions for the type-identity of cognitive states. My belief-that-p in 1991 is type-identical with my belief-that-p in 1995 just so long as both instances of belief-that-p have the same functional role. According to psychological reductionism I am nothing more than the connectedness and continuity of my psychological states. Thus, if a person is constituted by an inter-related collection of psychological states each of which is type-identical with an earlier person’s psychological states, the earlier and later person are identical.

            However functionalism does not specify whether a break in the spatiotemporal continuity of a person disrupts the type-identity of psychological states. The question does not normally arise, since functionalism is a thesis about the type-identity of psychological states, whereas the current issue is the continuing identity of an individual. The personal identity question and functionalism are related because, as noted above, an individual over time (or before and after teletransportation) can be viewed as a series of collections of type-identical psychological states. Functionalism might be held to support only Wide Reductionism, or also one of the Widest Reductionist views. Functionalism is not able to decide between these theories since it does not address the issue of spatiotemporal continuity; it only talks about type-identity between different individuals, or between earlier and later instances of the same spatiotemporally continuous individual. There is no obvious reason, however, why it is not compatible with the more liberal reductionist views. All discussions of functionalism assume spatiotemporal continuity since, in a world as yet lacking teletransporters and practicing Universal Immortalists, there are no instances of persons with spatiotemporal discontinuities.

            Since I want to defend CIWR, which allows identity to persist across spatiotemporal discontinuities, I will now leave functionalism aside. Different support is required to move beyond Wide Reductionism to the Widest Reductionist theories. This next stage is important because many people’s intuitions rebel when examples involve these discontinuities. Many of those who can accept My Brain Transplant and The Transporter are troubled by A Heroic Reconstruction, and cannot accept either Omega Point Resurrection or The Luckiest Person in the Universe as cases of continuing personal identity. The following series of cases begins with Parfit’s neurosurgeon example. Parfit uses that example to show the irrelevance of spatiotemporal continuity. To begin, here is Parfit’s story:

In Case One, the surgeon performs a hundred operations. In each of these, he removes a hundredth part of my brain, and inserts a replica of this part. In Case Two, the surgeon follows a different procedure. He first removes all the parts of my brain, and inserts all of their replicas. (p.474)

            In both cases the matter embodying mental items is different after the operation from what it is before. But we have seen that this does not concern us. In Case One there is spatiotemporal continuity between the old embodiments. One small part of the brain is removed at a time; the new part is attached to the brain and becomes part of it before another piece is removed and exchanged. Throughout Case One the same brain remains. In Case Two, if we require spatio-temporal continuity for physical objects then the same brain does not exist after the operation. The new brain is functionally identical to the old one but is a different brain.

            In both of the cases there will later be a person whose brain will be exactly like my present brain, and so that person will be psychologically continuous with me. “And, in both cases, this person’s brain will be composed of the very same new components, each of which is a replica of some part of my brain. The difference between the cases is merely the way in which these new parts are inserted. It is a difference in the orderings of removals and insertions. In Case One, the surgeon alternates between removing and inserting. In Case Two, he does all the removing before the inserting.” (Parfit, p.475.)

            The example can be made even more compelling if we specify that in both cases I am anesthetized before the operation. There is no difference in my experience between the two cases. Now, how can the difference in the order of insertions and removals be the difference between life and death? Why should it be vital that the new parts are temporarily connected to the old parts? If the order of insertions involved some further fact – perhaps a mystical force attaching only to gradual insertions and removals – then the order would be significant. But there is no further fact. The example can be varied by imagining an extremely rapid robot surgeon who is able to remove and insert neurons so fast that the new parts are connected to the old parts for only nanoseconds. This makes it even more implausible to think that the continued existence of the same physical brain is important.

            There is no significant difference between Case Two and The Transporter. Wide Reductionism and CIWR are therefore supported. The reliability of the cause of psychological continuity is irrelevant to the issue of identity. So long as all the parts of the brain are put together so as to produce psychological continuity we have maintained identity. The neurosurgeon might have changed her mind about finishing the operation. Perhaps after removing all the neurons she was told that I am an evil person and, being a utilitarian, she desisted from inserting the new neurons in order to rid the world of an undesirable person. Or perhaps she had a heart attack before inserting the new neurons. Many causes could stop the insertion process and thereby terminate the R-relation. Yet, surely, if the insertion does go ahead despite these uncertainties we have no reason to withhold the judgment that I have survived the procedure. It appears then, that this example supports our move at least as far as CIWR; any cause of continuity is sufficient. It need not be reliable so long as it does actually operate in any instance.

            It is hard to see any non-arbitrary way of stopping the move from Case One of the gradual insertions and removals to Case Two with its absence of spatio-temporal continuity or reliable cause. The lack of grounds for refusing to move from one case to another is the negative case in favor of CIWR. The positive case is that what we care about is the R‑relation (see the fourth argument, “Series-Persons”, below). Normally we are concerned about the continuity of the brain because normally such continuity is practically necessary to the maintenance of the R-relation. These cases just emphasize what we already have grounds to believe: that what matters is the R-relation and not its normal carrier.

            The neurosurgeon cases are just one example of a slippery slope that moves us irresistibly from intuitively plausible cases to those that are initially more difficult. Many other slippery slopes could be constructed. For example, if the opponent of CIWR tries to base his case on the reliability of the cause of continuity of the R-relation, we can vary circumstances in such a way that the continuity of the brain becomes unlikely. Brain continuity is never guaranteed: There are always strokes, falls, speeding vehicles, and falling objects. Suppose the conditions of the world changed so that the odds of brains surviving in one piece over the course of a day became very much smaller. I doubt that anyone would then say that the unreliability of my brain continuity meant that I was not the same person over time even though I was lucky enough to avoid brain destruction longer than most people. It would be tedious to construct more slippery slopes, so I will stick with the neurosurgeon cases. Once we see that it is the R-relation that we truly care about in these cases, we can see that many slippery slope moves could be described to undermine arbitrary restrictions on the causes of R-relation continuity.

SERIES-PERSONS

My elaboration of Parfit’s neurosurgeon example indicates the difficulty that will face anyone attempting to limit personal identity within a normal causal or spatio-temporal framework. In examining Nagel’s view that we are essentially our brains, Parfit adds further weight to the foregoing conclusion. He demonstrates that even if we accept Nagel’s claim that “I” and “me” refer to whatever actually makes possible my psychological continuity, we can still reach the Wide Reductionist View. The argument involves Nagel’s notion of a “series-person.” I will present Parfit’s argument briefly since I have nothing to add to it.

            Suppose that the previous arguments are unsuccessful and that I am essentially my brain and what matters to me is the continued existence of my brain. If this is what persons essentially are I, as a person, cannot choose to take a different view about what matters. But this leaves open another option. “While a person is, on Nagel’s view, essentially a particular embodied brain, a series‑person is potentially an R-related series of embodied brains.”[18] Nagel imagines a community in which everyone enters a Scanning Replicator once a year. The Replicator destroys the person’s brain and body and creates an exact Replica (except that it has not aged) who is R-related to this person. Nagel says that it would be rational for the series-persons of the community to use this Scanning Replicator and Parfit adds that such series-persons could live forever if they made back-up blueprints every day. Persons do not survive the Replicator but series-persons do.

            If Nagel is right that I am essentially my brain I cannot change my view about what I am. But I can henceforth use pronouns to refer to series-persons. The criterion of identity for persons is continuation of the brain but the criterion for series-persons is Relation R with any cause (i.e., it is the same as the criterion for persons if Nagel’s view is false). The words “I” and “me” now refer not to the person Max More but to the series-person whose present brain and body are the same as Max More’s brain and body. The previous sense of “I” and “me” will henceforth be expressed by “old-I” and “old-me.”

            Any activity that I carry out is being carried out by two different individuals if Nagel’s view is correct. Although the two of us both take the same actions at the same time we are distinct: If we go through a teletransporter then old-me, the person, would be destroyed, but I the series-person would continue. The series‑ person was not brought into existence by the invention of the concept. He has existed for as long as old-me, the person. Unless teletransportation, uploading of my mind into a computer, etc., become possible within my lifetime, I, the series-person, will also cease to exist at the same time as old-me.

            Supposing Nagel’s view to be true, what matters for old-me is the continued existence of my present brain. Despite this we can believe that what really matters is relation R, not brain continuity. Relation R is what matters for me, the series-person. Why speak and think as a series-person rather than a person? Consider another concept of Nagel’s: The day-person. Essential to the existence of a day-person is an uninterrupted stream of consciousness. Sleep or anaesthesia is death for a day-person. The concept of a day-person applies to reality but carves up reality into unimportant boundaries. We care about relation R and not the noninterruption of consciousness. The concept of a day-person is in this way inferior to the concept of a person. But then, if Nagel’s view is true, the concept of a person is inferior to that of a series-person in a similar way. The concept of a series-person appeals to what is more important—relation R.

            If Nagel’s view is false, the distinction between persons and series-persons is unimportant. Our criterion of identity will fail to cover many imaginary cases and some actual cases[19] (such as people with divided hemispheres). Since non-reductionism is false, questions about identity in those cases are empty. We can give answers to these questions by extending our criterion of identity, such as by making the criterion the non-branching holding of Relation R. On that criterion persons are series-persons and the distinction disappears. So, if Nagel’s view is false, the criterion for identity of persons allows persons to survive teletransportation. If Nagel’s view is true, persons cannot survive teletransportation but then series-persons can proclaim their existence and use pronouns to refer to themselves. Since the concept of a series-person would carve up reality in less arbitrary ways this would be an improvement.

            Consideration of series-persons therefore supports the Wide Reductionist View. It can easily also support the Widest View if we alter the example a little. In Nagel’s example, each year people enter a Scanning Replicator. In the altered example, we introduce an unreliable element of choice into the situation. Each person goes into the Replicator each year with the hope of being scanned, destroyed, and replicated in a de-aged body. However, the Scanning Replicators are controlled by a sect of technocrats who have taken it upon themselves to “improve society” by ending the existence of those they believe fail to meet their criteria for “good citizens.” They make these decisions after an individual has been scanned and destroyed but before they are replicated. If they decide against someone, their stored pattern information is randomized or overwritten. The process now is unreliable because it involves a choice. If the people use “I” to refer to series-persons, they will still survive the scanning and replicating process if those controlling it allow them to make it through. Altered in this way, the series-person argument supports not only the Wide View but also the Widest Reductionist View.

ARGUMENT BY UNDERMINING CONTRARY INTUITIONS

Despite the arguments and cases I have presented, doubt may linger over the case of A Heroic Reconstruction. Where the causal connection between phases of a person is highly roundabout and unreliable as in AHR, it may be hard to believe that CIWR is true. Even if the arguments I have given together constitute a strong case for this position, conflicting intuitions are likely to keep rising from the grave to haunt us whenever we forget about the arguments and thought experiments. As Hume noted, there are some philosophical views which go so much against the grain that when we leave the attentive reflection that convinced us of their truth we find ourselves once again feeling that a view is wrong even though we cannot support our intuition. (This can be just as true of physical theories such as relativity or the rejection of geocentrism.) I have found that my intuitions about the more difficult cases have gradually shifted. As the theoretical framework of the Widest View became embedded in my thinking my intuitive reactions to bizarre cases like AHR increasingly conformed to the theory.

            In this final section I want to further undermine intuitions that conflict with the theory by explaining why they exist. Intuitions arise because of well-entrenched assumptions formed often unconsciously or implicitly. If it can be shown that a certain intuition arises because it is normally well-grounded, but that it becomes over‑generalized, then we can undermine the reasonableness of forming beliefs on that intuitive reaction in cases outside its boundaries of applicability.

            For example, cognitive psychologists have demonstrated that humans typically use a few heuristic rules such as the representativeness and availability heuristics to make judgements in a wide range of cases. These heuristics work well much of the time but occasionally lead us to the wrong conclusions, as can be demonstrated by more sophisticated reasoning procedures such as probability theory. Once we have explained why we use the heuristics, and once we have defined their limited domain of applicability, we can deny any weight to intuitive judgements where they stray beyond their effective domain. Similarly, once we have Einsteinian and quantum mechanical physics—a physics well-grounded in theory and observation—we need not give weight to the Newtonian intuitions naturally derived from our normal experience of medium-sized objects with medium velocities.

            A common intuitive response when considering the role of causation in personal identity is that we only survive if the normal cause of our psychological continuity is maintained. For many educated people, this means they feel that the brain (or enough of the brain) must persist. For others, it means that that their “soul” must continue to exist, or perhaps that their soul must remain embodied. Especially for those many people who do not spend time thinking imaginatively about fictional or possible future scenarios, the very fact that, say, the brain is normally the cause of psychological continuity, is enough to wear deep grooves into their intuitions.

            In all real cases that we have ever encountered, and perhaps in all cases that we ever shall encounter, normal and direct causal connections are involved in the maintenance of psychological continuity and connectedness. Explanations of why I am much as I was a few minutes ago, and explanations of why I have changed somewhat from the way I was a year ago, always in fact involve causal explanations of a familiar kind. The relation R is, as a current matter of fact, embodied in brains and bodies. In the world as it is and always has been the standard physical causes we see in brains and bodies are practically essential to R-relatedness over time.

            Disruptive intervention in the causal chain, such as brain injury or destruction, or the degeneration of aging, terminates the R-relation. It is understandable that we are so concerned to ensure that the appropriate causes continue to operate. We know that if our brains decay or are destroyed then we go with them. This fact explains the attraction not only of the belief that normal, direct, and reliable causal connection is necessary to personal identity but also the appeal of the physical criterion of personal identity.

            However, as all the foregoing arguments have established, what really matters is R-relatedness. We care about psychological continuity and connectedness. If this can be secured by means other than the usual then we should be satisfied. Not so many years ago the idea of having your body disassembled, beamed across space, and reassembled would have terrified just about everyone. It would have been thought of as death. Now, very many people have got used to the idea by seeing it portrayed on Star Trek, by reading discussions of personal identity, or other sources, and the imaginary process is usually thought of as a means of travelling rather than as a means of dying. Many of these people, if they considered it, might therefore agree to the Wide Reductionist View, granting personal identity so long as there is some causal connection between the earlier and later persons. Unlike previous generations they do not think a normal cause is essential.

            The third case I described, A Heroic Reconstruction, will still upset many people, but the reason for this, is I think, that they are not used to considering such events. If someone were to produce a popular science fiction TV series based around the reconstructionist attempts of the Universal Immortalists many more people would gradually come to accept this as a way of surviving.

            Our intuitions are explained by our concern that our psychological connectedness and continuity be maintained. We are more uncomfortable about using a transporter than about normal continuity because we find it easy to imagine the transporter malfunctioning and leaving our disassembled atoms spread across space. We have no theoretical grounds for rejecting abnormal or unreliable causes of R-relatedness. Our concern is a perfectly reasonable practical concern. I will not place my trust in an unreliable cause if I can do better. If I have a choice of extending my lifespan by a proven and reliable medical technology, I will obviously choose that over the possibility that I might one day be reconstructed by Universal Immortalists. However, what I care about is continuing to exist, not the means of this continuation except in so far as this is a practical question. If psychological continuity is maintained by a more unlikely method I will not complain. Just as in the neurosurgeon case, I would much prefer that my neurosurgeon did not have a heart attack, but if she did and her colleague managed to bring about the same result then I would still have survived even though survival by that method was much less likely.

            We can conclude then, that while a normal, direct, or reliable causal connection between earlier and later persons may be practically necessary for them to be the same person, this is not conceptually or theoretically necessary. We care about the continuation of our personality, our goals, projects, values, memories, beliefs. Since normal, direct, reliable causal connections are standardly necessary to this kind of continuation we will also care about those kinds of connections instrumentally. But where standard kinds of causal connectedness turn out in a particular case to have been unnecessary to psychological continuity, we should realize that standard causal connections and psychological connectedness have come apart, and that it is the latter that should concern us. Having argued in favor of CIWR as the causal criterion to satisfy if we are to continue to exist, I will now turn to the question of when we cease to exist. Applying the Widest Reductionist View to the concept of and criteria for death shows up the deficiencies in the criteria currently used.

 

---

[1]Possibly also some accompanying physical features—see Chapter on “Technological Transformation and Assimilation.”

[2]Armstrong develops this distinction in “Identity Through Time” in Van Inwagen (ed), Time and Cause.

[3]For instance in Parfit (1971) he says we can “redescribe a person’s life as the history of a series of successive selves.”

[4]Hans Moravec, in Mind Children: The Future of Robot and Human Intelligence (recently supported in this thesis by Marvin Minsky), argues that the coming decades will see a transfer of human personalities into a synthetic hardware capable of running thousands to millions of times faster. If this happens, then a later self-stage only a year older in objective time could have developed so much that its earlier phases constituted a tiny portion of it.

[5] Parfit, 1984, p.206.

[6] Parfit, ibid, p.207.

[7] Parfit, ibid, p.208

[8] Nozick, 1981, p.39.

[9] Except that Nozick allows the possibility that features other than psychological ones may be constitutive of identity (p.69). I will examine the question of what features other than psychological we might include in the conditions for identity in the chapter “Technological Transformation and Assimilation.”

[10] Nozick, p.43.

[11] The Physics of Immortality. Doubleday, New York, 1994.

[12] Of course dualists and idealists may also believe the universe to be self-sustaining: They may be deists rather than theists.

[13]Kolak and Martin (1987).

[14]L. Weiskrantz, 1986, Blindsight. B. Libet, Behavioral and Brain Sciences, 1985,1987, 1989.

[15] This is a paraphrase of Nozick, p.36-37.

[16] Parfit provides several strong considerations against Nagel’s brain criterion for identity in  Parfit, 1984, S.93 and Appendix D.

[17] For a clear explanation of activation vector spaces and connectionist views of concept formation and learning see “On the Nature of Theories: A Neurocomputational Perspective,” and “On the Nature of Explanation: A PDP Approach,” in Paul M. Churchland, 1989.

[18] Parfit, p.289-290.

[19] Christine Korsgaard summarizes these cases in “Personal Identity and the Unity of Agency,” Philosophy and Public Affairs, Spring 1989, pp. 104-105.