In Choosing for Changing Selves, which I wrote towards the end of the 2010s, I tried to formulate a response to a worry that Edna Ullmann-Margalit, Krister Bykvist, and Laurie Paul, among others, had raised for the standard theory of rational choice. According to that theory, when we choose between different options that are available to us—whether or not to take an umbrella when we leave the house, whether to have pasta or curry for dinner, whether to become a parent, leave a relationship, or join a revolution—what rationality demands of us is determined by three things:
(i) the different ways we think the world might be—that is, the states of the world we consider possible;
(ii) how strongly we believe that the world is each of these ways—in the jargon, our credences in the various possible states; and
(iii) how strongly we desire or want or prefer the world to be in each of these ways—throughout, I’ll just talk of these conative attitudes collectively as our values.
However, as Paul and the others point out, all of these components change over time: we come to consider new ways the world might be—in the jargon, our awareness grows; we change the probabilities we assign to those different ways—sometimes because we receive new evidence, sometimes because we re-evaluate the evidence we already have, sometimes because we conduct some logical reasoning; and we change our values. So, when I choose now, which should I use? The states, credences, and values I have now? Those I have at some other time, past or future? Some amalgamation of all of these?
In the case of states of the world and credences, there are usually good reasons to stick with the ones your have now—I won’t go into these reasons in this post, but I’ve written a little about the more unusual cases where that isn’t quite true here and here. But the same is not obviously true of your values. There, we might well think that you should take into account the values you had at other times—in the book, as the title suggests, I talk of the values of your past and future selves; but the language of selves is intended as shorthand; everything could be rephrased in terms of what you value at different times.
Cases of what Laurie Paul calls personally transformative experiences furnish particularly striking examples of value change. While I currently value the aspects of my life afforded me by not having a child, I might predict that I would likely come to value those aspects less were I to become a parent. And so I might think that, while my current values, together with my current beliefs, lead me to prefer remaining childfree, my future values, should I become a parent, would lead me to prefer the life as a parent that I would then have. When I face the choice whether to begin the process of adopting a child, to which values should I appeal? If I am using standard decision theory, which values should I measure numerically to give my utilities, and then include in an expected utility calculation? My values now? Or my different possible future values—the same as my values now if I remain childfree, but quite different if I do adopt? Or some amalgamation of all of these?
In Choosing for Changing Selves, I argue for an amalgamation. The overall value you should currently assign to a particular possible state of the world—measured numerically to give what I call your global utilities—should be a weighted sum of the values of the different selves that comprise you in that state of the world—the values of each self measured numerically to give what I call its local utilities. So, in the state of the world in which I adopt and become a parent, I assign weights to the values of each of my past selves, I assign weight to the values of my current self, and I assign weights to the values of each of my future selves. After discounting each of these values by the assigned weights, I add them up to give my overall value for that state of the world, and then I plug that into standard decision theory in the usual way.
But now a question arises: how should I set the weights? What norms govern how I do this? In the book, I suggested there were two sorts of norms: first, norms of rationality; second, norms that are like moral norms but directed towards other parts of oneself, or other of one’s selves, rather than other people. So, for instance, one norm of rationality might be that, if you discount future selves according to a function of how far in the future they lie, so that selves further in the future receive less weight than those more imminent, and you use the same function to do this at all times, it should be what economist’s call an exponential discounting function. Economists, for instance, have argued that we are rationally required to do this because doing otherwise makes us vulnerable to sure losses. And, for instance, a rather minimal norm of the other, quasi-moral sort might demand that you give at least some positive weight to any given self—unless the values of that self are somehow beyond the pale, morally speaking, in which case you can and must give them no weight at all. While a rather stronger and more controversial norm would be one that obliges you to give a certain level of positive weight to the values of a past self, or series of past selves, who sacrificed a great deal of well-being in their time in order that you be in a better position at the current time.
I spoke of this other sort of norm as being like a moral norm but directed at oneself, rather than simply being a moral norm, because I’d bought a claim common in modern moral theorising that says that moral norms govern your behaviour towards other people, while only prudential norms of rationality govern your behaviour towards yourself. But, reading some of the recent literature on duties and obligations to the self—specifically, Alison Hills’ ‘Duties and Duties to the Self’, Paul Schofield’s Duty to Self, Daniel Muñoz and Nathaniel Baron-Schmitt’s ‘Wronging Oneself’, and Collis Tahzib’s response to Schofield—I realised that this is historically anomalous, and that most earlier moral theorists considered it both possible and unremarkable that there should be moral norms concerning one’s treatment of oneself. So now I’m happy to think of as moral the norms that I sought to govern the weights one should give to the values of other selves.
However, categorization aside, this doesn’t help answer one of the central questions in this area: how do the norms governing how I should weight the values of others of my selves relate to norms governing how I should weight the values of other people? In Choosing for Changing Selves, I often introduced the idea that there was a specific norm governing weightings for other selves by drawing an analogy with a similar uncontroversial norm governing weightings for other people. For instance, I tried to argue that we sometimes ought to give a certain amount of weight to the values of our past selves by drawing an analogy with (very defeasible) filial duties to give weight to the values of parents. But I never said that I thought the analogy was perfect and the norms governing the two exactly parallel.
So I was very intrigued by the argument in Muñoz and Baron-Schmitt’s paper that exactly such a parallel holds in the case of rights. That is, they argue that, if I have a right against you that you not do X to me, then, in exactly parallel situations, I also have a right against myself that I not do X to me. Of course, at first, this might seem absurd: surely I have a right against you that you not pinch my arm, but equally surely I have no such right against myself? But Muñoz and Baron-Schmitt argue that I do have exactly that right against myself. But, just as I might consent to let you pinch my arm and thereby change the normative facts so that there is no moral prohibition on you pinching my arm, so I can do something like consent to let myself pinch my arm and thereby again change the normal landscape so that I’m not morally in error when I do pinch it.
Their paper is essentially a defence of this position against some of the objections one might naturally raise, and they do an excellent job of providing this defence. They focus mainly on cases in which they claim one’s current self has a right against that very same current self not to do X to them. This is natural, since it is in these cases that such rights have the initial whiff of paradox: I, now, have a right against myself, now, not to pinch myself now—are you sure? But I think in fact these are not the cases that pose the greatest challenge to their theory. The greatest challenge comes from cases in which there is something we intuitively think it is permissible for me, now, to do to a reasonably far future self—perhaps myself in a couple of months’ time, a couple of years’ time, or a couple of decades’ time—that we do not think it is permissible for me to do, now, to another person. There are, it seems, burdens that it is acceptable for me, now, to impose on myself in a couple of months time that it would not be acceptable for me to impose on any other person. Perhaps I spend all my money this weekend on an enormously lavish celebration, and thereby subject my future self in a month’s time to penury; or perhaps I get a cosmetic procedure that carries a reasonably high risk of physical harm to myself in ten years’ time. It wouldn’t be acceptable to do the analogous thing to my partner. I shouldn’t spend all of our shared earnings, for instance, on something that benefits me alone; nor should I spend all my money on myself alone if my partner is financially dependent on me; and I shouldn’t get some cosmetic procedure myself that creates a reasonably high risk of physical harm to my partner. But, you might think, depending on the enjoyment I gain from the celebration this weekend and depending on the misery I endure from the penury in a month’s time and depending on how long its lasts, that it is an acceptable trade-off for me to do with my future self; and depending on the pleasure the cosmetic procedure affords me now and depending on the nature and risk of the physical harm to my future self, that is also an acceptable trade-off. Can Muñoz and Baron-Schmitt account for such cases in their framework?
I’m not so sure. For how could my future self consent to the penury my current self imposes on him, and thereby waive the right he has not to suffer the harms and indignities of immiseration? Of course, you might worry that a related problem arises when my current self violates his own rights, for while my current self clearly could consent to my current self pinching his arm, he in fact does not—I never find myself saying to myself, “Yes, sure, go ahead and pinch me; that’s fine.” But Muñoz and Baron-Schmitt have a response to that worry. They appeal to an increasingly popular observation in the literature on consent that there are many ways in which a person can consent to something without explicitly stating or gesturing that they consent; or, perhaps better, there are ways in which something that would be impermissible becomes permissible in the way it would were consent given, but which don’t involve consent at all—consent changes the normative facts in a striking way, but it is not the only way in which such change can be effected. And they appeal, in particular, to John Gardner’s argument that, when something is an essential part of a joint project on which both parties have embarked together, both parties have consented to doing that thing: “In an ideal tango,” Muñoz and Baron-Schmitt write, “you do not need to ask your partner for permission before each twirl. The fact that you are mutually engaged in the dance, responsive to each other’s movements and in touch with each other’s wants, is enough.” Now, while it is rather eccentric to say it, I have plausibly embarked on a joint project with my own current self, and I am certainly in touch with his wants, and so actions that are an essential part of whatever that project might be gain automatic consent; or, again, perhaps better, they are made permissible in exactly the same way they would be made permissible were consent given. Perhaps it’s hard to see how pinching my own arm is essential to the joint project on which I’ve embarked with myself, but if we describe that project at a fine enough grain, so that it simply involves doing what I in fact want to do, I think we can make out the idea.
Let me grant that this all works for those cases in which (i) my current self violates the rights of that very same current self, but (ii) we say doing so is morally permissible. But how might it be extended to those cases in which (i) my current self violates the rights of my future self, and (ii) we still say doing so is permissible? If Muñoz and Baron-Schmitt are to again appeal to Gardner’s argument that one can waive one’s rights by embarking on joint action with another, the question arises: how does my future self embark on this joint action, and thereby waive his right? He is yet to exist, and as such he is as yet unable to embark on anything.
Nonetheless, you might think my future selves are automatically engaged in joint action with my current self. Christine Korsgaard suggests something like this in her early response to Derek Parfit’s views on personal identity and their implications for morality.
[S]uppose that a succession of rational agents do occupy my body [as Parfit suggests]. I, the one who exists now, need the cooperation of the others, and they need mine, if together we are going to have any kind of a life. The unity of our life is forced upon us, although not deeply, by our shared embodiment, together with our desire to carry on long-term plans and relationships. (113)
Korsgaard goes on to say that this is a little misleading because it suggests that each momentary self only values what happens to that self at that time, and they negotiate with other selves to live a life that obtains as much as possible of what each self values. But of course many of the interests of my current self involve those future selves and not only as tools for the satisfaction of the current self’s preferences.
In order to make deliberative choices, your present self must identify with something from which you will derive your reasons, but not necessarily with something present. The sort of thing you identify yourself with may carry you automatically into the future; and I have been suggesting that this will very likely be the case. (113)
Now, recall the concern I raised for Muñoz and Baron-Schmitt: we often permissibly burden future selves in ways we could never permissibly burden other people. We might appeal to Korsgaard and say the perspective from which we deliberate requires us to view ourselves as engaged in joint action with our future selves; and so, following Gardner, this changes the normative facts: our future self waives their right not to be so burdened by being engaged in this joint activity with our current self, where that joint activity requires that burdening.
But I don’t think that can work. One problem is that everything Korsgaard says about the relationship between my current self and my future self also holds for the relationship between me and my family, my group of friends, my neighbours, my colleagues at work, those with whom I volunteer, and so on. Consider the following rewriting of the passages from Korsgaard I just quoted:
I need the cooperation of my family, friends, and neighbours, and they need mine, if together we are going to have any kind of a life. The unity of our life is forced upon us, although not deeply, by the geographical space we share, the resources on which we must jointly draw, and the material and psychic goods we can provide to one another. In order to make deliberative choices, you must identify with something from which you will derive your reasons, but not necessarily with only yourself. The sort of thing you identify with may carry you automatically among the people with whom your share the spaces of your life; indeed, this will very likely be the case.
Now Korsgaard recognises this analogy:
If personal identity is just a prerequisite for coordinating action and carrying out plans [which is what Korsgaard has been arguing previously], individual human beings do not have to be its possessors. We could, for instance, always act in groups. The answer to this is surely that for many purposes we do; there are agents of different sizes in the world. Whenever some group wants or needs to act as a unit, it must form itself into a sort of person—a legal person, say, or a corporation.
And later she writes, suggesting that a state is a good analogy for a person:
A state is not merely a group of citizens living on a shared territory. We have a state only where these citizens have constituted themselves into a single agent. They have, that is, adopted a way of resolving conflicts, making decisions, interacting with other states, and planning together for an ongoing future.
But the problem with Korsgaard’s view is twofold:
First, even when all that she describes as constituting a state has happened—when the people of that state have adopted a way of resolving conflicts, making decisions, interacting with other states, and planning together for a future—still we need the practice of consent because being part of the same state doesn’t deprive an individual member of their rights against certain behaviour by others. And so, however strong the analogy between the relationship between a person and their component selves, on the one hand, and the relationship between a state and its citizens, on the other, it doesn’t follow that my decision now can assume the consent of another self at a later time.
Second, there is no way for my current self, my past selves, and my future selves to constitute ourselves into a single agent in this way. We cannot together adopt a way of resolving conflicts, making decisions, interacting with other agents, and planning together for an ongoing future. After all, we never co-exist. A person comprised of different selves is not like a state comprised of citizens all of whom exist together in a way that enables them to negotiate their plan for living together; it is instead like the succession of generations that comprise a state extended over time. But the current generation living in a state is not permitted to burden a future generation in the way we often take a current self to be permitted to burden a future self—it is not permissible for my generation to pollute, despoil, and befoul the planet in a way that will burden future generations. The current generation in a state has long-term goals that require the cooperation of future generations for their fulfilment; and they might require burdening some of those future generations; and in some sense the generations are involved in a joint activity, even if the future ones have not actively embarked on that joint activity with the current generation, but are rather thrown into it, much as I rarely embarked on the joint activity of a rugby game when I was at school, but was rather thrown into it. But none of this means that the future generations automatically waive their right not to be burdened by the current generation in certain ways. The problem with Korsgaard’s picture is that it gives too much power to the current self to deliberate from the point of view of an extended agent that includes a future self who has no power to refuse to be included in that agent and who might not have common cause with the present self at all—as the literature on transformative experience makes clear, my future selves might have a radically different conception of how they would like the life of which they are a part to go.
This is one reason I don’t think Korsgaard’s view of the unity of agency can help us to answer the concern about Muñoz and Baron-Schmitt’s view. My current self and my future self haven’t embarked on a joint project in such a way that certain rights of my future self are automatically waived; my future self isn’t there to negotiate the way in which they’ll execute this joint project and the way they’ll distribute the burdens in the way that the tango partners or the members of a state are all there to carry out these negotiations with one another; my future self may be part of the same joint project as my current self, and he might have no choice in this matter, it having been imposed on him by sharing the same body with me, but it is also true that my family and neighbours are part of a joint project with me, and they might have no choice in this matter, it having been imposed on them by the accidents of birth and location. In neither case does this mean that any rights are waived automatically.
But there is another reason Korsgaard’s view can’t help us, namely, that there are cases in which we do wrong ourselves. That is, there are cases in which I do something that burdens one of my future selves in a way that is morally impermissible. As I mentioned above, the current literature on duties and obligations to self makes clear that this was the standard view prior to the twentieth century. And it’s plausibly the folk view now. Paul Schofield points to the example of Monty Brogan, the protagonist of Spike Lee’s 25th Hour, who is staring down a long prison sentence for selling drugs. After railing against the city he lives in, the characters he interacts with there, the corrupt police, and so on, he turns on himself, castigating his past selves for their actions that brought him here: “You had it all, and you threw it away.” Brogan recognises that his past selves wronged his current self: they behaved in a way that risked the extremely burdensome situation in which he now finds himself. And yet those past selves, we might suppose, acted according to the plan for the life they share with the current self who is now reprimanding those past selves in the mirror. We might even imagine that those past selves of Monty Brogan considered the risk of prison, knew it would fall not on them but on the future selves, and went ahead all the same. That is, those past selves, when deciding whether to make another drug deal, did deliberate from the point of view that Korsgaard imagines we must, namely, one from which they were reconciling the conflicting wants of the different selves of Monty Brogan at different times, and acting with a view to their shared life. But, all the same, he wrongs himself.
This highlights what I think is the central problem in this area; the problem I grappled with in the second half of Choosing for Changing Selves, but for which I gave no principled, general answer, instead working piecemeal through some cases based on intuitive judgment. It’s possible to wrong others; and it’s possible to wrong ourselves. It’s also permissible sometimes to burden others; and it’s permissible sometimes to burden ourselves. But there are many cases in which I’m permitted to place burdens on myself that I’m not permitted to place on others. Intrapersonal morality is less demanding than its interpersonal counterpart. The symmetric view that Muñoz and Baron-Schmitt offer can’t account for that unless it’s easier to obtain consent for certain behaviour from my future selves than it is from the current selves of others. But it’s very unclear why that would be—indeed, we might think it’s the other way around, since the current selves of others are present and may be asked for their consent, while my own future selves are not. But similarly the view that intrapersonal tradeoffs are always morally permissible providing they’re rational also fails to account for this fact. For instance, it isn’t clear that Monty Brogan’s past choices weren’t prudentially rational, for the risk he took might well have maximized his expected lifetime utility. But nonetheless the past selves that made those choices wronged his current self by imposing an unacceptable risk of extreme burden on him—a burden that he did, in the end, have to bear, though we might think that it would have been wrong to take the risk whether or not he’d had to bear it in the end.
It’s tempting to think that what makes the difference here is that Monty Brogan’s current self does not endorse the decision that his past selves took. So his past selves chose against the preferences of his current self. But it can’t be required for morally permissible behaviour that affects your future selves that you do exactly what they would have you do. Take the case of the person who spends all their money on a wonderful celebration and immiserates their future self. The future self won’t endorse that choice, but it doesn’t seem to become immoral because of that. And, if it does seem immoral to you, lessen the proportion of the person’s wealth spent on the celebration and thereby lessen the burden on the future self enough that it is morally acceptable, but would not be morally acceptable if imposed on another person, even one with whom one’s life is deeply entwined, such as a partner or friend.
So I think the puzzle still stands: how do we reconcile the fact that there are burdens it would be prudentially rational but morally impermissible to impose on one’s future self, and yet there are burdens it’s morally permissible to impose on your future selves that it isn’t morally permissible to impose on others? How do we explain the asymmetry between the intra- and interpersonal case without saying that there are no moral demands in the intrapersonal case?
In the next couple of posts, I want to explore two ways we might try to answer this. The first asks whether we can adapt Judith Jarvis Thomson’s thought experiment of the famous violinist from the context in which she used it to shed some light on the matter; and the second asks whether Susan Wolf’s style of argument in her ‘Self-Interest and Interest in Selves’ might help us.
I feel that the discussion of "rights against oneself" does more to illuminate the difficulty with the notion of rights that is relied on (roughly, some notion of natural rights).
If you view rights as being created by a state or similar body, the problem goes away. I have a right to stop you pinching my arm in that, if you do it, I can call the police and have you charged with assault. In this sense, I can't have rights against myself. At most, I can conclude that I can't trust myself and hand over my rights to a trusted third party.
Obviously, rights of this kind are not necessarily the rights we should have, and there can be conflict between different systems of rights (for example, those spelt out in international agreements, and those actually recognised by the states in question). But this still seems more coherent than thinking about natural rights in a way that gives us rights against ourselves
This is a great article!
One thing I’m not so sure I agree on is that we should care about our future preferences. Our current preferences care about our future life, but I’m not so sure that should be true of those future lives’ preferences as well… Sometimes we should because this “change in preference” is actually just a deeper reflection on what we actually would want under more idealized circumstances, but I’m not sure we should actually give weight to REAL different preferences.
Shouldn’t it be in our interest for our preferences not to change? Shouldn’t we not want to do something in the future that we now hate?
This is a different article that I liked that seems related and is (imo) worth checking out:
https://www.lesswrong.com/posts/Kbm6QnJv9dgWsPHQP/schelling-fences-on-slippery-slopes