1. Two sorts of narrativism

A growing number of philosophers propose narrativist accounts of personal identity. The nature of personal identity, they say, has something to do with the stories we tell about ourselves. Narratives can be ‘identity-constituting’.

There are many different questions of personal identity (Olson Reference Olson and Edward2016, §1), and not all narrativists address the same one. Some are concerned with the so-called ‘characterization question’, others with the ‘persistence question’. Some never make it entirely clear which they mean. But there are two very different sorts of narrativism.

The characterization question expresses what nonphilosophers typically mean when they speak of personal identity. It asks for someone’s most characteristic or distinctive properties: those that ‘define her as a person’ or ‘make her the person she is’ (cf. Glover Reference Glover1988, 109f.; Schechtman Reference Schechtman1996, 1f., 74). They are typically mental and behavioral properties: commitments, preferences, beliefs, character traits, and the like. It is common to refer to these properties collectively as a person’s ‘identity’. To have an ‘identity crisis’ is to become unsure of one’s most characteristic properties. If, in an anguished moment, I ask, ‘Who am I?’, I want to know what sort of person, in some deep and fundamental sense, I am.

Narrativist accounts of characterization say that the properties making someone the person she is have to do with narratives (cf. DeGrazia Reference DeGrazia2005, 82f., Schechtman Reference Schechtman1996, 94, Shoemaker Reference Shoemaker2009, 89–91). An answer to the characterization question will take the form of an autobiographical story. Who am I? Well, I am someone who has done or experienced this and that, which has led, in certain ways, to my having such-and-such commitments, preferences, character traits, and so on. What makes me the person I am is the stories I tell.

The characterization question asks for certain mental and behavioral properties of someone. The persistence question, by contrast, asks what it takes for someone to continue existing from one time to another, or what makes a past or future being you. It is the question of personal identity over time.¹ An answer will be a completion of the formula

where the ‘ = ’ sign expresses numerical identity. Lockeans fill the blanks with something to do with psychological continuity: what makes a past or future being you is for the mental states she is in then to be causally dependent, in a certain way, on those you are in now, or vice versa. Animalists fill them with some sort of brute-physical continuity having no psychological component: what makes a past or future being you is its being the same biological organism that you are. Narrativist accounts of persistence are presented as an alternative to these familiar views. They say that what makes a past or future being you is something else, to do with narrative connections.

The view is no straw man. Jeanine and Robert Schroer, for example, explicitly propose a narrativist answer to ‘the question of (diachronic) personal identity’–that is, ‘the question of what makes a person (x) at time t1 and a person (y) at time t2 the same person’ (Reference Schroer and Schroer2014, 446). Anthony Rudd says that ‘the concept of narrative…is central to thinking about the identity of persons across time.’² And although Marya Schechtman claims to be primarily concerned with characterization (Reference Schechtman1996, 2, 94; Reference Schechtman2014, 100), she sees her view as addressing the worry that ‘psychological continuity theories…do not allow for the persistence of a single, experiencing subject’:

As you and I are temporally extended (that is, persisting) subjects of experience, this appears to say that we persist by virtue of ‘identity-constituting narratives’. Actions or experiences necessarily belong to the same subject of experience, and hence the same person, whenever the right sort of narrative is in place. Some condition involving narratives is at least sufficient for a person to persist from one time to another. Similar statements abound in her work.³ Both Schechtman and Rudd present their accounts as rivals to Lockean and animalist views.⁴ If they were not answers to the persistence question, they would be entirely compatible with those views and not rivals at all.⁵

Our interest is in narrativist accounts of persistence. (The characterization question will reappear in the final section.) We believe that they face grave objections. This is in addition to those they share with psychological-continuity theories. The principal objection to psychological-continuity theories arises because they rule out our being biological organisms: no sort of psychological continuity is either necessary or sufficient for an organism, including a human animal, to persist (Olson Reference Olson1997, 94f.). This raises the awkward question of what sort of non-organisms we might be, and why a human organism cannot itself be a person. Persistence narrativism has the same consequence, as no condition involving narratives is necessary or sufficient for an organism to persist either. But we will ignore this point.

2. Formulating persistence narrativism

What would a narrativist account of persistence look like? The proposals offered have not always been clear or explicit. The most obvious suggestion is that the right sort of narrative connection is both necessary and sufficient for a person to persist. A past being is you just if you now have narratives of the right sort identifying you with her as she was then. A future being is you just if the narratives she has then identify her with you as you are now. So:

The bits in parentheses are to avoid a well-known transitivity problem. Suppose that, owing to some pathological condition, the earliest event figuring in your current self-narratives is your first day at university, but that on that day the student had narratives extending back to nursery school. Without the parenthesis, the proposal would imply that you are the student and the student is the toddler at nursery, but you are not the toddler, violating the transitivity of identity (which is a theorem of standard first-order logic with identity). Our formulation avoids the problem by implying that you are the toddler, as you now relate to her as she was then by a chain of narrative connections. It may also need a ‘non-branching’ clause to deal with cases where two later beings have narrative connections to a single earlier one (see Schroer and Schroer Reference Schroer and Schroer2014, 463). We will set this aside.

To understand this proposal we need to know two things. First, what is the right sort of narrative? What makes a narrative ‘identity-constituting’–the sort that figures in persistence? And second, what is it to have a narrative?

Most narrativists say that the stories have to be autobiographical: you must be their narrator and central protagonist. They have to be told by you, from your point of view. And they must be about what you take to be your own life (Schechtman Reference Schechtman1996, 124, 142–145; Reference Schechtman and Hutto2007, 167; Rudd Reference Rudd2009, 61). (We will revisit this claim in §8.)

They must also play an explanatory role, saying something about how you came to have your most characteristic properties–those relevant to the characterization question–and to undertake the momentous actions in your life. An identity-constituting narrative must be ‘a story of how the events in one’s history lead to other events in that history’ (Schechtman Reference Schechtman and Hutto2007, 160).

Accuracy, however, cannot be required. The narratives may relate events that never happened, or give false accounts of real events. Otherwise the view would be circular. Suppose one of your self-narratives had it that you won the sack race at the St Brutus’s Primary School sports day in 1963. In order to verify this, we should have to know whether the winner of the race (supposing it actually took place) was you or someone else, which is a fact about your persistence. We should have to know which past being was you before we could apply the theory that was supposed to tell us this.

Having a narrative is understood dispositionally (Schechtman Reference Schechtman and Hutto2007, 161; Rudd Reference Rudd2009, 62; Schroer and Schroer Reference Schroer and Schroer2014, 455). You need not consciously formulate the story. You must only be in some way disposed to tell it in certain circumstances.

To simplify formulations, we will abbreviate the phrase

as

So the proposal is that a past or future being is you just if he or she is then narratively continuous with you as you are now. Our persistence consists entirely in narrative continuity: nothing more is needed, nothing less will suffice. Call this pure narrativism.

3. Pure and impure narrativism

Pure narrativism is hopeless. Suppose you become convinced that you are Shakespeare reincarnated. It’s part of the story in terms of which you understand your life that you were born in the 16th century, acted at the Globe, wrote the plays and poems attributed to Shakespeare, and so on. You have a detailed account of how these events affect your current character and the principal events of your life since your latest reincarnation. It’s all a delusion got from history books, but you have forgotten this and take yourself to have genuine memories of events from the poet’s life. The story might be perfectly coherent and entirely sincere. It identifies you with Shakespeare, making you narratively continuous with him as he was then. It follows from pure narrativism that you really are Shakespeare.

The obvious response is to add a further condition. Your having an appropriate narrative connection to a past or future being is necessary for him or her to be you, but more is needed. Schechtman proposes a ‘reality constraint’: though identity-constituting narratives need not be entirely true, they must be consistent with ‘the basic character of reality and…the nature of persons’ (Reference Schechtman and Hutto2007, 163; see also Reference Schechtman1996, 119–130). If one of the facts about ‘reality and the nature of persons’ were that we cannot be reincarnated, this would solve the problem: once we’re dead and our remains are dispersed, we cannot come back into being.

The trouble with this suggestion is that once again it appeals to a claim about personal identity over time. Whether we can be reincarnated depends on what it takes for us to persist, which is what persistence narrativism was supposed to tell us. We should have to know something about our persistence conditions already, by other means, before we could work out the theory’s implications.

A better thought is that the narratives must be based on real first-person memories, so that no event could feature in such a story unless one remembered it (Schechtman Reference Schechtman and Hutto2007, 167; Schroer and Schroer Reference Schroer and Schroer2014, 463). We could then appeal to the fact that no one can remember (or ‘quasi-remember’) experiences had by someone dead and buried, as this blocks the causal connections that need to hold between an experience and a genuine memory of it (Shoemaker Reference Shoemaker, Shoemaker and Swinburne1984, 81–85). This is not a fact about our persistence: it does not turn on what it takes for a person to continue existing. It would rule out your being Shakespeare because you have no first-person memories of events from his life.

Call this further constraint C, and call the view that our persistence consists in narrative continuity together with C impure narrativism.

4. When do we begin?

Impure narrativism has its own difficulties. The most obvious is its startling consequences about the beginning of our lives. We do not come into the world already outfitted with self-narratives. First we do and undergo things; only later do we reflect and work them into stories. We cannot form identity-constituting narratives without the cognitive skills needed to think systematically about past events, which few of us acquire before age four. And owing to childhood amnesia, few of us can retrieve memories from our earliest years. So by the time you begin to think about your life in narrative terms, you will probably have no memories of events that took place before age two at least (Peterson, Grant, and Boland Reference Peterson, Grant and Boland2005, 626). And in order to avoid the Shakespeare problem, narrativists need to say that an event can figure in your identity-constituting narratives only if you remember it.

Suppose the first event that figures in your self-narratives occurred at age two. And suppose, as impure narrativism tells us, that you extend only as far into the past as those narratives do: you could not have come into being before the first events that feature in those stories. It follows that you did not exist before your second birthday. The implication is not merely that you were not a person then--the persistence question is not about what it is to be a person or a nonperson--but that you did not exist at all. The child who was born, learned to walk, and began speaking in simple sentences was not you, but a numerically different being. In fact the child could never have become a person: if she had, she would have persisted without narrative continuity during her first years, contrary to impure narrativism. It follows that if an ordinary child dies of pneumonia at eighteen months, this neither kills a person nor prevents any being from becoming a person. It only prevents a person from coming into existence, much as contraception does. Most of us will find this bemusing.

Now standard psychological-continuity theories have a similar consequence (Olson Reference Olson1997, 73–76, McMahan Reference McMahan2002, 44–46). If our persistence consists in psychological continuity, we could not have existed before we had any mental properties at all: as five-month-old foetuses, for instance. A foetus of that age persists without any psychological continuity. It follows that it could never become a person. The death of a five-month-old foetus neither kills a person nor prevents any being from becoming a person. It can only prevent a person from coming into existence. This is something that psychological-continuity theorists have grudgingly accepted. But no one, narrativists included, thinks we come into being at age two.

5. Suicide by narrative discontinuity

Impure narrativism has surprising consequences about the end of life too.⁶ It would make it possible for us to destroy ourselves without incurring any physical injury. We need only give up our self-narratives–that is, eradicate the disposition to tell certain stories. There would then be no narrative continuity between ourselves as we are now and ourselves as we were previously. If narrative continuity is necessary for us to continue existing, as narrativism says, this would literally be fatal. Breaking off all narrative continuity would not merely end our existence as people, but destroy us altogether. There would be no need of tall buildings or sleeping pills.

Not that this would be easy to do. We have no direct voluntary control over our self-constituting narratives. It is nearly impossible to remember events from one’s past without imposing some sort of narrative on them. It’s as natural as breathing, and requires no more conscious effort.⁷ But we could retain the disposition to think narratively without being disposed to tell any particular story. This may require a lengthy programme of mental exercises, hypnosis, or mind-altering drugs, but it does not seem psychologically impossible, for a time at least. The result would be complete narrative discontinuity. This could happen even if full psychological continuity is maintained. (Though psychological continuity may be necessary for narrative continuity, it cannot also be sufficient. Otherwise the view’s consequences would be no different from those of psychological-continuity theories. It would not be an alternative to standard accounts of our persistence, but merely Lockeanism in new clothing.)

Or at least we could do this if we ceased permanently to have self-narratives. But old habits may creep back. At some point the resulting person may devise new stories about the events of your life–events which he or she would remember as well as you did. If the new narratives stretched back to the time when the original stories were extinguished (as they almost certainly would), this person would be you. Giving up one’s self-narratives would be a difficult and unreliable method of suicide.

Or rather, its effectiveness would be limited to special cases. Imagine that you fall into the hands of bad people who are going to torture you to death. If you could eliminate your self-narratives for the duration of the torture, you really would cease to exist and spare yourself the agony. (Secret agents may find the skill worth cultivating.) Unhappily, though, your departure would create a new conscious being in your place–someone just like you only lacking self-narratives. She would take herself to be you (as she would have all your memories), and mistakenly believe that the suicide attempt had failed. So even if deliberate narrative discontinuity spared you from torture, someone else would suffer it.

To our knowledge, narrativists have never considered these points.

6. Shifting narratives

We have been discussing the troubling consequences of impure narrativism, the view that a being existing at another time is you just if there is narrative continuity linking you as you are now with her as she is then and condition C is met. But the view is also internally inconsistent.

Imagine that someone, call him Sigmund, acquires his first self-narratives at age four and that they reach back a year. Narrativism says that our existence extends exactly as far into the past as the events featuring in our identity-constituting narratives do. So if Sigmund constructs his first self-narratives in 1860, and they stretch a year into the past, then he must begin to exist in 1859, three years after his birth (a surprising claim, but we’ve been through that already). In adult life, however, he may develop self-narratives reaching back another year. And again, narrativism says that his existence extends just as far into the past as his identity-constituting narratives do. He must, then, have begun a year earlier.

The example is hardly fanciful. In fact narrativists often emphasize that our narratives are subject to revision (e.g. Rudd Reference Rudd2009, 65). But it entails two incompatible claims: that Sigmund begins to exist in 1859 (as that’s how far his first self-narratives reach), and that he begins in 1858 (as that is the extent of his later self-narratives). Call this the problem of shifting narratives.

No doubt narrativism could be fixed to avoid the problem. If the self-narratives you have at different times in your life extend back to different dates, it might be the ones extending to the earliest date that determine the beginning of your existence. So even if the self-narratives you develop in your youth go back only a year, you might exist earlier than that owing to stories you construct later. And if you should lose the self-narratives that reach back to a certain date and retain only stories beginning later, you might still exist at the earlier date owing to self-narratives you had previously. We could formalize the proposal like this:

We might call this proposal complex narrativism. Whatever virtues it may have, however, it does nothing to address the objections raised elsewhere in this paper. And it is unattractive in other ways. If when we begin depends at least partly on our self-narratives, why should it depend on only some of them? And why only on those that reach back furthest? The narratives you have early in life are not in any way shallow, inauthentic, or otherwise defective simply because later ones extend further. They may be stories of just the sort that narrativists take to determine the length of our existence. They would do so, according to complex narrativism, if no narratives you had at other times reached back further. Narrativism’s central thought is that what makes an earlier being you is the narrative connections you now have to that being as he or she was then (perhaps together with condition C, which we assume to hold). So if those connections now extend back only to a certain time, that’s how far your existence ought to reach. Complex narrativism allows that none of your current narratives make any difference to when you began, simply because you later develop more far-reaching ones. You could never infer the date of your beginning from the self-narratives you have at a given time: they could always be overruled, as it were, by stories you have at other times. Few narrativists will welcome this.

7. Retroactive self-creation

Here is a final difficulty. All the narrativist theories we have considered imply that when we come into being depends on how far into the past certain self-constituting narratives extend. We exist at a past time only if we have self-narratives now–or at some time, anyway–identifying ourselves as we are at that time with a being existing at the past time. And we create these narratives after the events featuring in them: first we do and experience things; later we work them into stories.

How far back the stories reach depends on which events from our childhood we remember and how much interest we take in them, among other things. They may extend only as far as events occurring at age four. Or they may go back to age two or three. It all depends on what happens later: on what stories we take up as we reflect on our past.

But if we come into being at the time of the earliest events in our self-narratives, then the date of our beginning will depend on what we do later. Had our later narratives differed in how far they reach into the past, we should have begun at a different time. We literally bring ourselves into being in early childhood by developing narratives in later life. (Those narrative acts are not sufficient for us to begin–our parents’ contribution was also needed–but they are necessary.) We begin to exist when we do because of what happens later. This is a case of backwards causation.

Metaphysicians disagree about whether backwards causation is ever possible–in extraordinary cases such as time travel, for instance (Lewis Reference Lewis1976; Mellor Reference Mellor1998, 125–35)–but none of them believes that it happens in common and familiar situations. Yet persistence narrativism appears to imply that backwards causation is an everyday occurrence. No one could come into being without it. If affecting the past were as easy as that, we ought to wonder why we cannot now bring it about that such baleful events as the destruction of the library at Alexandria, the second world war, or the election of certain presidents never happened. For those whose lives have gone badly wrong, there ought to be a remedy even better than suicide: bringing it about that one never existed. This is hardly what narrativists have in mind when they speak of our unique ability to ‘create ourselves’ (Glover Reference Glover1988, 131; DeGrazia Reference DeGrazia2005, 89f.).

8. Social narrativism

We have been exploring the idea that some condition involving narrative continuity is both necessary and sufficient for us to persist. It implies that we come into being years after we are born and that we can end our lives simply by changing our way of thinking. It leads to inconsistencies about our beginnings, and entails that we literally bring ourselves into being as children by later acts of narration. No rival account of our persistence has these consequences, and no one wants them. We will devote the remainder of the paper to attempts at salvaging persistence narrativism.

One thought is to give up the idea that identity-constituting narratives must be autobiographical. Other people’s stories about our lives might be relevant too. Schechtman writes,

Our persistence might be determined not only by our own stories about ourselves, but also by those that parents, friends, and biographers tell about us. People who are never able to tell their own stories may persist entirely because of this. Suppose, then, that a past being is you just if someone now has narratives of the right sort identifying you with her as she was then (or there is a chain of such narrative connections). And a future being is you just if the narratives that someone has at that future time identify her with you as you are now (or, again, there is a chain of such connections).

Call this social narrativism. It has the virtue of avoiding the implication that we come into being years after our birth. Even if infantile amnesia prevents our own self-narratives from including events of the first two years of life, those of others can fill in the gap. Your parents’ story of you may extend even to the moment of conception.

But it’s hard to see many other advantages in social narrativism. It may imply that giving up your self-narratives would not by itself end your life: the stories of others could secure your survival. But it would still be possible to destroy yourself without physical harm by getting everyone else to expunge their narratives about you as well.

Nor does it solve the problem of shifting narratives. Suppose, again, that Sigmund’s mature self-narratives date his beginning to an earlier time than his juvenile stories do. If his own stories sufficed to determine his persistence, this would have inconsistent consequences about when he began. Perhaps the stories of others might somehow cancel out this inconsistency. Still, it could happen that all those whose narratives determine his persistence say first that he began at one time and later that he began at another, with the same inconsistent consequence. And this is only one instance of a more general problem: What happens when the narratives constituting someone’s persistence disagree? Suppose the stories of one faction date Sigmund’s beginning to 1856 and those of another to 1858. They cannot both determine his beginning. Which ones count? The trouble threatens all narrativist accounts of persistence, but allowing multiple narrators exacerbates it.

Social narrativism also threatens to revive the Shakespeare problem. This time condition C is no help: it requires you to remember the events that figure in the relevant narratives, and social narrativism does away with that requirement. Nor does the proposal avoid backwards causation. We should still come into existence because of stories told later. It makes no difference who tells them.

9. Weak narrativism

A more promising thought might be to say that some narrative condition is sufficient for us to persist, but not necessary. (The corresponding view that it is necessary but insufficient has no attraction.) We could survive without narrative continuity. We actually did so during infancy, before the earliest events that figure in our self-narratives. We may do so again at the end of our lives, after our narratives have been destroyed by senile dementia. But if a being is, at some past or future time, narratively continuous with you as you are now (and C holds), she is you. Call this weak narrativism. (It may be what Schechtman intended in the passage quoted in §1.)

Weak narrativism does not imply that we could end our lives by giving up our self-narratives, and is consistent with our having existed as infants before the earliest events in those narratives. These consequences follow only from the claim that narrative continuity is necessary for us to persist. Likewise, it avoids the problem of shifting narratives by not implying that we extend only as far back as our self-narratives do. And for the same reason, it does not imply that we bring ourselves into being retroactively by creating self-narratives later in life: for all it says, we may have existed before any of the events that figure in those stories.

But it is unattractive in other ways. It says, in effect, that our persistence consists sometimes in narrative continuity and sometimes in something else. What makes you a certain pre-narrative toddler would presumably be some sort of non-narrative continuity: perhaps the psychological continuity of Lockean views, or some sort of brute-physical continuity as most animalists say. And if your narratives were destroyed by a deliberate effort of the sort described earlier (or perhaps by senile dementia), you would again persist by virtue of non-narrative continuity. Yet during the ‘narrative period’ of your life, from the time of the first event that figures in your self-narratives until those narratives are lost, only narrative continuity counts, and the other forms of continuity are irrelevant.

Weak narrativism goes against the narrativists’ central claim: that what makes a being existing at another time you is a matter of narrative continuity. In this respect it resembles the view that psychological continuity is sufficient but not necessary for us to persist (Olson Reference Olson1997, 81–85). The thought is that we survive as foetuses by virtue of brute-physical continuity, but once we acquire the relevant mental properties, this becomes irrelevant and only psychological continuity matters–yet if we lapse into a persistent vegetative state, our survival consists once again in brute-physical continuity. This is a view that almost no one holds: all Lockeans we know of say that some sort of psychological continuity is both necessary and sufficient for us to persist.

Nor does weak narrativism answer the persistence question. It does not complete the biconditional formula set out in §1, but only its right-to-left conjunct. As an account of our persistence, it is radically incomplete. It tells us nothing about how long we existed before the earliest events in our self-narratives. For all it says, we might come into being at the time of those events, between the ages of two and four. Or it could be at birth, or at conception, or perhaps earlier still. You could even be Shakespeare reincarnated. Again, condition C is no help: it was proposed as a necessary condition for our persistence, and weak narrativism gives only sufficient conditions. Because your narrative continuity with Shakespeare would not satisfy C (you can’t remember events from his life), weak narrativism does not entail that you lived in the 16^th century. But it doesn’t rule it out either. In fact, for all weak narrativism says you could be Shakespeare even if you have no narrative connections with him as he was then. If we want to know when you began–what makes you the infant or foetus or Elizabethan poet you once were, if any–we must look to an account of persistence not involving narratives.

Analogous questions arise about what happens when we lose our self-narratives. Does it destroy us? Do we survive as long as there is psychological continuity? Do we carry on as lifeless corpses till only dust remains? Again, weak narrativism is silent. An answer to these questions requires an account of our persistence involving non-narrative continuity. But if we need such an account to deal with the non-narrative periods of our existence, why bother trying to combine it with a narrativist account? Why not simply accept that we persist in all circumstances by virtue of non-narrative continuity?

Weak narrativism is messy, counterintuitive, and unprincipled. And it still appears to require backwards causation. Although it does not imply that we retroactively bring ourselves into being, it entails that what suffices for us to persist changes in early childhood from non-narrative to narrative continuity. The change takes place at the time of the earliest events that feature in our identity-constituting narratives. Both the occurrence and the time of this change depend on what narratives we have later on. Acts of storytelling in our middle years are responsible for changes to our persistence conditions in early childhood: later facts or events cause earlier ones.

10. Saving narrativism

If this is the best narrativists can do in giving accounts of persistence, they’re wasting their time.⁸ But we have given no objection to narrativist answers to the characterization question. It may still be that what makes you the person you are has to do with narratives. Narrativists can avoid the problems we have discussed by confining themselves to characterization.⁹ A view making no claims about persistence runs no danger of implying that some of us are Shakespeare reincarnated, that we could destroy ourselves just by changing the way we think, or that we can affect the past. It is even consistent with our being animals (DeGrazia Reference DeGrazia2005, 76).

The fact that narrativists have never considered these objections suggests that many of them have not adequately distinguished the characterization from the persistence question. Their loose formulations, which often leave uncertain which one is at stake, only reinforce this diagnosis. It would improve the narrativist literature to no end if all parties made it completely clear what question they were trying to answer.¹⁰