Self Matters

  • Marie Guillot (University of Essex)
  • Lucy O'Brien (University College London)


We argue that relating to myself as me provides, as such, a reason to care about myself: grasping that an event involves me, instead of another, makes it matter in a special way. Further, this self-concern is not simply a matter of seeing in myself some instrumental value for other ends. We use as our foil a recent skeptical challenge to this view offered in Setiya (2015). We think the case against self-concern is powered by unwarrantedly narrow construals of three key notions. One is the notion of a first-personal way of relating to oneself. A narrow account of the first person in terms of special epistemic relations to oneself makes it easy to overlook a source of non-instrumental reasons of self-concern, located in the special relation a subject has to herself as agent. Two is the notion of what it is to be a reason. And three is the notion of self-concern itself. We show that the skeptical case rests in part on a slide towards neighbouring but distinct notions of egoism and selfishness. We also argue that Setiya’s notion of self-love, offered to capture the pre-theoretical intuition of self-concern, cannot do it justice.

Keywords: self-concern, the first person, agency, non-instrumental reasons for concern, pre-suppositional reasons, self-love, selfishness, egoism

How to Cite:

Guillot, M. & O'Brien, L., (2023) “Self Matters”, Ergo an Open Access Journal of Philosophy 9: 28. doi:



Published on
31 Mar 2023
Peer Reviewed

1. The Intuition of Self-Concern

That something will happen to me seems to give me a reason to care about it in a special way. Moreover, I seem to care about it because it will happen to me. This intuition is induced by the following kind of case:

TRIAL: Suppose I took part in a randomized trial where just one of ten participants was given a tablet containing a new drug against diabetes, and the others a placebo. Unfortunately, halfway through the trial, the scientists discover that the drug has unwanted side-effects: the person who took the drug has a very high chance of becoming paralysed. Learning this is a reason for me to feel sorry for the unlucky participant who took the active substance. Learning that I took the active substance, however, provokes a reaction that is of a different kind.1

What would be one’s response in such a situation? Here is how Kieran Setiya describes it:

I shudder in dread, much more distressed than I was before. Self-concern is addressed to my wellbeing from the first-person perspective: to my well-being considered as mine, not just that of one among many. The fact that [someone] will be harmed may be a reason for me, as for anyone, to care; the fact that I will suffer goes beyond this. (2015: 446)

So, what is the nature and basis of my special concern that something bad—or worse—will happen to me? Prima facie, the rational source of my concern about my future is a distinct kind of source from my concern about another participant’s future, even supposing that one of the other participants is my friend: I care about my future because it is mine; my caring about their future is due, as Wollheim puts it, to some other attitude I have about them. Wollheim, in his attempt to characterize the difference between the case that involves my friend, and the case that involves me, describes the former case as making an ‘impact’, but the latter as inducing a ‘tremor’ (1984: 237). One hits me hard, and might harm or hurt me; the other makes the very ground shake.

In this paper, we want to defend the view that the fact that something will happen to me gives me a reason to care about it in a special way: that there is a structural and rational distinction between ‘tremor’ and ‘impact’. We want to argue, in particular, that there are first-personal reasons for self-concern: something happening to me is the reason for this particular kind of concern. Moreover, we want to defend the view that this kind of concern comes from the way in which I relate to myself first-personally—not just in relating to this particular human being, but in relating to her as me—and that it is not conditional or instrumental:2 it is rather a pre-supposition of a self-conscious creature having any other kind of concern. In arguing for this view, we will use as our foil a recent skeptical challenge to it offered by Kieran Setiya in his paper ‘Selfish Reasons’.

Setiya, as we saw above, recognises the intuition of self-concern. In order to fix a target for his more critical project, Setiya tries to capture the intuition of self-concern in the following principle of SELF-CONCERN:

SELF-CONCERN: The fact that an event will benefit or harm me is a reason for me to want, or not to want, that event to happen. This reason derives from the effects of the event on my well-being, not its effects on anything else. And its force as a reason turns on its first-person character. (2015: 445).

SELF-CONCERN construes concern of the kind we started with in terms of having reasons to want turning on their first-person character. Despite being concerned to articulate, and make sense of, the intuitive concern I seem to have about what befalls me in particular, a central purpose of Setiya’s paper is to argue that SELF-CONCERN is false. The paper argues that I have no reason for having an interest in my own fate, based on the fact that it is mine, and which does not turn on its effect on anything else: I have no non-instrumental first-person reasons for self-concern. This, of course, means that Setiya needs to look elsewhere in an attempt to make sense of the intuitive concern we started with. He tries to make sense of self-concern not on the basis of first-personal reasons but on the basis of ‘self-love’ which he takes to be ‘a disproportionate concern for the interests of the beloved’ (2015: 469). We can take the principle that captures Setiya’s account of self-love to be:

SELF-LOVE: The fact that an event will benefit or harm me is a reason for me to want, or not to want, that event to happen. This reason derives from love of myself, which, like all love, is a rational, albeit, disproportionate concern for the interests of the beloved. It is rational in being grounded in the fact of a person’s humanity. It is disproportionate in that it is stronger than the concern we are required to have for strangers, as such, while not being grounded in first-personal, or other personal, reasons.3

We will argue that Setiya has not established that I do not have a reason for having a non-instrumental interest in my own fate based on the fact it is mine. This will offer us the opportunity to show that we do have the resources to offer a non-instrumental account of the reasons behind self-concern. We will also argue that SELF-LOVE will not suffice for capturing the kind of self-concern manifested in the particular sort of dread—the ‘tremor’—we started with. Clarifying the failures of self-love as an account of self-concern will enable us to show, as Wollheim has argued, that self-concern is non-arbitrarily distinct from other-concern, and is non-optional. It is rather constitutive of what it is to live the life of a self-conscious rational agent.

2. Setiya’s Argument against SELF-CONCERN

Setiya’s argument against SELF-CONCERN has, roughly, two elements. The first involves a conditional claim to the effect that, if there were first-personal non-instrumental reasons for interest in our well-being they would have to flow from the way we relate to ourselves first-personally. The second is to admit that there may be first-personal reasons that flow from the way we relate to ourselves first-personally, but deny that they are non-instrumental reasons.

The key argument that gets us to the denial of non-instrumental first-person reasons is what we can call Setiya’s ‘why care about the object of IMMEDIATE KNOWLEDGE’ argument. The argument involves the following components.

First, we have the claim of IMMEDIATE KNOWLEDGE.

(IK) IMMEDIATE KNOWLEDGE: When I think of myself in the first-person, I do so in virtue of standing to myself as the object of immediate knowledge, knowledge that is non-inferential, nontestimonial, and immune to error through misidentification relative to the concept of myself that figures in the self-ascription of beliefs. (The kind of immediate knowledge involved here includes ‘knowledge of what I believe, but also knowledge through proprioception, perceptual perspective, and intentional action’ (2015: 461).)

IK is a constitutive claim about the nature of first-person reference; it is a claim about what it is in virtue of which I can self-refer first-personally. IK would only be a claim about first-person reference, set apart from the claims being made about the nature of self-concern, were it not for Setiya’s second principle:

HARMONY: Where a subject thinks about an object, in virtue of standing in a particular relation to it, that relation must accord with the rational significance of the corresponding thoughts.4

Now ‘accord’ is not a particularly helpful guide here to what Setiya must have in mind, given that he takes IK and HARMONY to be jointly sufficient to justify the conclusion that SELF-CONCERN is false. On the face of it, ‘accord’ can signal a number of relations. It may be something as weak as compatibility, or something as strong as entailment. It may be a symmetric or an asymmetric relation. If A accords with B, can B accord with A, must B accord with A? We will assume here that accord means something like determines, or explains. Setiya’s argument barely gets off the ground unless we take it in this sort of way. Claiming that the epistemic relations invoked in IK must be compatible with allowing a rational significance to thoughts based on SELF-CONCERN, does not constitute an impediment to claiming a rational significance of such thoughts. If IK turned out not to be a source of such significance, their rational significance would only have to come from some source, distinct from, but compatible with, IMMEDIATE KNOWLEDGE to comply with the constraint set by HARMONY. That is not Setiya’s conclusion.

The third step of Setiya’s argument is the step that earns the argument the name we have given it: the ‘why care about the object of IMMEDIATE KNOWLEDGE’ argument. Setiya claims that the epistemic relation that secures first-person thought—immediate knowledge—is not a basis for special, first-person based, concern:

Why care so much about the one you know first-hand, without the need for inference, whose beliefs you can access in a special way? The epistemic relation that secures first-person thought is not a basis for special concern . . . (2015: 467)

Just because I know a person, me, more directly, without inference, in a special way, why should I care particularly for her? From a certain epistemic privilege in our access to the object of first-person thought, Setiya sees no reason to infer any normative consequences, any consequences for the regard with which we should treat this object. He allows that being related to myself via the epistemic relations invoked in IK might give me a reason to care about myself, in so far as it presents me with the fact of my humanity. My humanity is a reason to care about myself, as it is a reason to care about anyone, and indeed everyone. So, the fact that an event will benefit or harm me is a reason for me to want, or not to want, that event to happen, and this reason does derive from its effects on my well-being, not its effects on anything else. However, its force does not turn on its first-person character. Therefore, SELF-CONCERN is false.

3. A Constructive Diagnosis

Something like SELF-CONCERN is very intuitive. However, it is also very hard to articulate the grounds for it. The normal kinds of reasons to care about something appeal to the properties that that thing has. But this explanation falls short in this case: here we have a case in which we seem to have special reasons to care, without those reasons corresponding to special properties of that which is cared about. There is no characteristic that I have, that is unique to me, and that grounds the particular concern I have for myself. And while, according to Setiya, there is a relation (captured by IK) I bear uniquely to myself, a merely epistemic relation of this kind is unsuited to motivate special concern for its target. Our hope is that a consideration of Setiya’s argument, and how we might resist it, will help us better work out what the grounds for self-concern are, and what kind of reason a first-person reason must be. How might we resist his conclusion? There are, as far as we can see, four options:

  • (A)

    Reject Harmony.

  • (B)

    Argue that IK does not give the right account of first-person thinking.

  • (C)

    Argue that IK does give first-personal non-instrumental reasons.

  • (D)

    Deny that the Principle of SELF-CONCERN properly captures the intuition behind self-concern.

Let us start with (A). Should we reject HARMONY? Suppose we accept that something like HARMONY is generally true. The rational significance of our thoughts about an object, then, generally flows from the relation we must have to this object in order to form thoughts about it. We might still wonder whether first-person thinking provides an exception. There is a well-established tradition of exceptionalism about ‘I’ in this regard. We know that first-person thinking is hard to explain in the way we explain other forms of thinking about objects, and we might think that our expectations in relation to explaining the rational significance of our first-person thoughts should be tempered as a result. John Campbell is well known for arguing that the ‘patterns of use of a concept’—where concepts are thought constituents—usually accord with the reference rule for the concept: we can explain the patterns of use by appeal to the reference rule. However, the first-person, he claims, is an exception. The first-person is governed by the self-reference rule and it:

does not provide one with a way of finding either the bases for making first-person judgments or the consequences that one draws from those judgments. (Campbell 1994: 110)

If Campbell is right—perhaps because the self-reference rule is the right reference rule for ‘I’—then Setiya’s argument runs to ground on HARMONY. Exceptionalism about the first-person will mean we should not expect HARMONY to apply here.

But perhaps we need to distinguish the rules of reference operative in thoughts from the conditions that enable a subject to think thoughts with such rules of reference operative? (As Heck 2002 and Dickie 2015 do in relation to demonstratives, and Salje 2016 does in relation to the first-person.) Suppose one thinks that the self-reference rule is the right way to explain reference determination in the case of first-person thoughts, but also thinks that certain further conditions are needed to enable a subject to think first-person thoughts with the reference so determined. What sort of conditions might those be? Well, presumably, only those determined by those relations in which a subject can stand to herself without prior identification, that is, relations such as those invoked in IK.

Let us, then, clarify HARMONY so as to make it clear that it applies not only if one takes the relevant conditions to be those that determine reference, but also if they act in other ways as conditions on a subject’s ability to think first-person thoughts. It is easy, in the context, to read Setiya’s condition ‘where a subject thinks about an object, in virtue of standing in a particular relation to it’, narrowly, as a claim about conditions on reference. We suggest restating HARMONY so that it specifies what is intended by ‘accord’, and makes it clear that we can take a wide reading of the conditions on first-person thinking, so that claims of first-personal exceptionalism are less troubling.

HARMONY*: Where a subject thinks about an object, in virtue of standing in a particular relation to it—due to that relation playing a role either in reference determination or in thinking a thought so determined—that relation must determine the rational significance of the corresponding thoughts.

It may be that HARMONY* is false even on the broader reading we offer. Perhaps the rational significance of first-person thinking is not related, in the way demanded, to the relations to ourselves that enable or constitute such thinking. We, however, find it very plausible that, if there are non-instrumental first-person reasons for concern about ourselves, the explication of them will be determined by the way of thinking operative in first-person thought. And since the interest for us of Setiya’s argument is not merely a defensive one, but an exploratory one, we will accept that something like HARMONY* is true.

This takes us to our second option:

  • (B)

    Deny that IK embodies a satisfactory account of first-person thought?

There is a puzzling aspect of Setiya’s discussion. He is committed to HARMONY—and in particular, he is committed to the idea that non-instrumental first-person reasons, if there be any, must be made intelligible through an account of the first-person way of thinking. He also, very vividly, expresses the intuition that leads to the formulation of the principle SELF-CONCERN. He declares failing to find a basis for SELF-CONCERN in his favoured account of first-person thought—embodied in IK. But why, then, is he not more flexible about what might be involved in first-person thinking? Why does he not explore the possibility that we could adapt and improve the account of first-person thinking in such a way that it would deliver a basis for self-concern? Struck by this puzzling intransigence, we—in this paper—are going to work backwards and ask: given the demands of HARMONY*, and the intuition behind SELF-CONCERN, what view of first-person thought might be adequate? What ways of being related to ourselves could plausibly play a role both in first-person thinking, and in explaining how we have special reason to care about ourselves?

In his account of first-person thought Setiya identifies certain special epistemic relations: knowledge of what I believe, knowledge through proprioception, knowledge through perceptual perspective, knowledge through intentional action. However, it is an important fact about the way I am related to myself, that it is not only that I know the individual I happen to be in these special ways. I also affect her in special ways.5 I am related to that individual’s actions in a way quite unlike the way I am related to others’ actions, and not just in an epistemic sense.6 The individual whose actions and states I immediately determine in acting is the individual that is me.

Setiya himself considers whether an agential element is missing in his account of first-personal thought, and whether IK as it stands risks ‘obscuring the role of agency in fixing its object’ (2015: 467). He goes on to ask whether, ‘when I think of myself in the first-person, I do so in virtue of standing to myself as the object of both agency and immediate knowledge’ (2015: 462).

So, let us look to adding a role for agential relations in first-person thought. Perhaps alongside IK we could propose the following principle:

IMMEDIATE EFFECT: When I think of myself in the first-person my doing so depends on my standing to myself as the object of immediate effect: if in acting I immediately change something, that thing is me.

Now consider a parallel question to Setiya’s ‘Why care’ question: ‘Why care so much about the one you affect immediately?’ But we would, as authors of intentional change, care about the agent, and state of the agent, that is both the source and the result of those changes. If the animal that I am is required to enact, by changing, all of the things I do, how can I not care about the state of that animal? Indifference would be unintelligible, unless indifference to my actions were intelligible. But being indifferent to one’s agency would be to fail to be an agent at all. The fact of my agency may not imply any particular evaluations, but it must imply that such are possible.

Is this enough? Setiya thinks not—he claims that while such an appeal to agency might give us first-person reasons for self-concern, they give us only instrumental self-concern:

The relation of agency has practical significance, but the value it confers on the agent is instrumental. I should matter to myself as the basic means to efficacy in the world, the source of intentional action. It follows that I have good reason to care about my own well-being. But against Self-Concern, this reason derives from the role of my well-being as a means, not as an end. (2015: 468)

So, Setiya agrees that it is important to keep track of how things are with me when I deliberate about what to do, given that I am the agent, but this is only because being the agent means I am the means of achieving my goals: I am my own tool. We agree that one can sometimes be one’s own tool, but want to resist the idea that this is the primary way of relating to oneself in action. In what follows, we argue that non-instrumental reasons of self-concern can be grounded in agential relations to myself, because these are primarily relations, not to the instrument of my actions, but to their site and to the determiner of their ends.

3.1. The argument from non-objectuality

Notice first that there is an important distinction to be made between a human animal being the agent, and a human animal using themselves as an instrument in agency. This is a distinction between oneself acting and acting on or with oneself. The latter, as well as the former, is required for a human animal to use themselves as a means, but the former alone is required for being the agent, and is, in our view, the basic case.

Let us give an illustration: perhaps someone who suffers paralysis in one arm could decide to use her paralysed hand as a tool for typing, moving it with her other hand. But her typing in this way would be very different from the way she typed when she used the other, non-paralysed hand, to hit the keys. It would not be the basic mode of typing. We might describe the former way of typing, typing with my hand, and the latter way my hand typing. Using one’s hand, or any part of one’s body, as a tool, is an available mode of action, but it is not, we think, the basic mode.

The ability to think a first-person thought is (by IMMEDIATE EFFECT) based on standing in a relation to oneself as agent (inter alia). Indeed, this much Setiya concedes. But to relate to oneself as agent is not to relate to oneself as the instrument of one’s action, as one might relate to a wooden spoon, or as the imagined subject might relate to her paralysed hand. In these cases, the objects used as tools are not objects of immediate change; their changes are mediated by the changes immediately effected in the acting hand. Thinking a first-person thought is not based on relating to oneself as to an instrument, because the basic relation I stand to myself in acting, appealed to by IMMEDIATE EFFECT, is not as an instrument.7

Instrumentality comes with the possibility of an assessment of something being good for some end, and selected as the right means. But in the context of my own action, there is no question of choosing myself or someone else as the means of my action. Being the subject of action is structurally different from (even when it is concurrent with) being the instrument acted upon.8

But how exactly does this point help us resist the claim that the only first-person reasons for self-concern are instrumental reasons? The first step is to identify what we take to be an objectual condition on instrumentality.9 Put simply, for me to value, and hence view, any thing10 as a means, or an instrument, is for me to view it as an object of use to some ends. This constraint can be put as follows:

Objectual constraint on instrumentality: For some O to be valued as a means or as an instrument by S is for it to be thought by S that O is good/useful to φ with, to bring about some end.

And in the first-person case:

Objectual constraint on instrumentality (first-person): For some O to be valued as a means or as an instrument by me, in the context of my own action, is for it to be thought by me that O is good/useful for me to φ with, to bring about some end.

So, given the objectual constraint on instrumentality, to value or view oneself as a means is to view oneself as an object O that it is good or useful to φ with. But an agent does not, in general, take such an attitude to herself. She does not think of herself as some thing that it would be good or useful to φ with. In the basic case an agent thinks it would be good/useful to φ, simpliciter—perhaps using some object O, usually distinct from herself, in doing so.

This gives us a negative argument against the suggestion that first-person reasons are instrumental in the way suggested by Setiya. In thinking that it would be good or useful to φ, I do not thereby think of myself as an object to act upon, or with, because I do not think of myself primarily as an object at all. I think of what speaks in favour of φ-ing, and what against, of whether to φ, or not. I do not think about what speaks in favour of me φ-ing, or whether to φ with myself, or someone else.11 The place of the agent in such thinking thus fails to satisfy our objectual constraint on instrumentality.12

However, thinking about the role of the agent in determining what to do also provides a place to start in giving a positive argument for the claim that there are first-person reasons for action which are not instrumental, and not due to my being a means to some end. We want to argue next that there are reasons of a kind—primary reasons—that give us non-instrumental reasons for self-concern that turn on their first-person character. These reasons derive from my role in my own agency, not as a means, nor indeed as an end, but rather, as we might put it, as a beginning or source of my capacity to set ends. My capacity to so much as set ends, and in turn secure the means (including where—in non-basic cases—this means is me), depends on the nature of the agent I am, and that gives me a primary reason to be concerned with what I am. How might that positive argument go?

3.2. The argument from the agent as a source of concern

Our positive case for non-instrumental reasons of self-concern turns on the thought that there is room in the structure of my own action for more than just the means, and the end. A further role is the source of value: that which both chooses the ends, and assesses the correct means to achieve them. As agent, I may sometimes be the means of my action. We have also argued that I am always one of my ends (in the sense of results I bring about), since whatever I immediately change in acting is me (by IMMEDIATE EFFECT). But it is easier still to see that I am always also the source of the action: the setter of ends and the selector of means.

It is a point made familiar by the Kantian tradition that whoever occupies that third role has a particular standing. To be that which can see something as valuable (whether as a means, or as an end) is to have a special kind of primary value, and a reason to value oneself as such.13 While this point is not uncontroversial,14 we find it very compelling, and we think it can be accepted without committing to the entire framework of Kantian ethics, as evidenced by its wide appeal beyond this school of thought.15 In any case, the idea that a valuer must value herself need not be construed as a Kantian claim about the metaphysics of value-setters. In our view, its plausibility comes, rather, from its status as a presupposition of the notion of a subject valuing anything. For any appraisal to stand as such, the appraiser must value not just the object of appraisal, but the appraising and whoever is performing it, too, since the authority of the appraiser plays an essential role in giving the appraisal what weight it may have. (It would be absurd for the Olympic winner to value their gold medal but have no faith in the discernment of the judges.) It is internal to the idea of evaluation that in valuing anything, I also value the grounds on which it is valued, and the valuer herself.

But how exactly might my role as a source of value yield non-instrumental reasons of self-concern? Our argument goes as follows:

  1. For an object O to be valued as an instrument by S (and so for O to be something S is concerned with as a means) is for it to be thought by S that O is good/useful to φ with.

  2. For me to have instrumental value in my own eyes, I would need to be the O that is valued as good/useful for φ-ing with.

  3. My fundamental place in the relation of concern as means, however, is as that which values things as good/useful as means, relative to an end, not as the means that are valued as good/useful.

  4. To be that which values things as good/useful as means, relative to an end, is to have a special kind of primary value: it is to be a thing that matters in itself, whatever else matters. (Cf. Kant 1785/1998.)

  5. So if, in first-person agency, I relate to myself primarily as something that other things are good/useful for as means, the concern introduces a kind of primary concern which is a non-instrumental concern.

  6. My concern for any object O as good/useful to φ with implies self-concern in this primary non-instrumental sense.

There are a number of clarifications to be made that will help to bring out the role of this argument in providing first-person, non-instrumental, reasons.

First, we should note that this argument does not give us what we might call ‘categorical goodness dependent’ non-instrumental reasons as the basis for our self-concern. The dialectic inherited invites us to choose between grounding non-instrumental first-personal reasons for self-concern either on recognition of (i) a categorical goodness property I instantiate that provides me with first-person reasons for self-concern, or on recognition of (ii) myself as an object able to secure some categorical goodness property. It is then pointed out that (i) cannot give first-personal reasons—recognition of categorical goodness dependent properties must be recognition of them as universal reasons, and (ii) cannot give non-instrumental reasons—concern rooted in recognition of myself as an object able to secure some categorical goodness property is instrumental concern. The conclusion is drawn that there can be no first-person non-instrumental reasons. However, that conclusion is only secure if all non-instrumental reasons are indeed based on instantiations of goodness properties. The argument above suggests that setting up the dialectic in this way overlooks a source of first-person reasons. In particular, in the role each has with respect to herself in being that in relation to which she can choose things as good, or useful, as means in her actions.

Suppose an individual A φs with an object O. Suppose A takes O to be good or useful for φ-ing with, and suppose that she has categorical goodness dependent reasons for φ-ing. A must be such that she recognises or determines O as good or useful for φ-ing with. In recognising or determining O as good or useful for φ-ing with, she exercises her capacity to relate to herself as that which things are good and useful for relative to an end—in this case the end of φ-ing. She has instrumental reasons to be concerned with O, and she has non-instrumental categorical goodness related reasons to be concerned with φ-ing.

What reason does she have to be concerned with herself? There is some sense in which she need have no concern with herself at all: she determined whether to φ, found that there were decisive non-instrumental reasons for φ-ing, and that there were instrumental reasons for φ-ing with O, and φ-ed with O. She need not attend to or contemplate what reasons she has to be concerned with herself in order to rationally φ with O. However, if she understands her role in the structure of reasons on the basis of which she does φ with O she will appreciate that:

  1. There is no distinction between determining whether to φ, and determining herself to φ.

  2. Taking O as good or useful for φ-ing, is for herself to take O as good for useful for φ-ing with.

  3. To be concerned with φ-ing as an end, on the basis of the reasons she has, and to be concerned with O as good for useful for φ-ing, on the basis of reasons she has, is to be concerned with herself φ-ing, and taking O as good or useful for φ-ing.

To appreciate this is to appreciate that the force of a reason, if it is to function as a reason in the determination of action, is inseparable from her determination of it as a reason. This is a general point about the relation between reasons and the first-person, perhaps made most clearly in recent discussion by Burge. Consider an agent being moved to act in recognition of a non-first-personal reason such as an individual’s being human. Such a reason is only able to function as a reason that guides an agent in thought and action, in so far as it is determined to be a reason by the agent determining what to do. As Burge puts it, ‘this function [of reasons] operates only through the reasoning of individuals’ (2000). An agent could not be rational, appreciate the role she has in relation to reasons, and not have first-person reasons for self-concern.

It may be objected that while the considerations above show that the capacity to be a reasoner, to act on the basis of reasons, is indeed a source of value and concern, it has not been shown to be a source of reasons that turn on their first-person character. An agent who appreciates the role she plays in determining the force of a reason, can appreciate that others play the same role,16 and can appreciate the non-first-personal value in being a reasoner, that can be put alongside the values of being human, beautiful, knowledgeable etc. We think that the considerations above do indeed show that that the value of being a reasoner is inseparable from the other categorical goodness dependent reasons. It might not be quite right to treat it as ‘one more reason alongside’ the others, but we can set that point aside. The relevant point for our purposes is that the considerations above, besides giving us a reason to be concerned with reasoners in general, also give a distinct reason for each reasoner to be concerned with themselves, first-personally. After all what they appreciate is the role of the first-person, themselves, in the determination of their reasons for thought and action. A subject’s appreciation of the general value of being a practical reasoner, and the constraints it puts on their thought and action, depends—as all reasons do—on their determination of it as a reason that constrains their thought and action. And if they understand that then they must also understand not just the value of being a reasoner in general, but their—first-personal—value in being a reasoner taking the value of being a reasoner as a reason.17

Even if the grounds we have articulated for self-concern have consequences for what rational reflection will put me in a position to grasp about the concern owed to other agents—concern grounded in the recognition that they are also determiners of what it is good or useful to do, and with what—the relation I have to the ground for my self-concern implies a deep structural difference between these two forms of concern, and the roles they play in the life of a rational agent. My standing in relation to my determination to φ now with O is presupposed: the other’s determinations to φ now with O are always candidates for my sceptical assessments.18

If we are right, reasons for self-concern dependent on our capacity for practical reason come in at least two kinds—two kinds that are easy to run together. We recognise both (a) universal non-reflexive reasons to value our role as reasoners. These will give reasons that everyone has a reason to act on in every case; and (b) universal reflexive reasons for each to value her own role as a reasoner. These will give reasons that everyone has a reason to act on in their own case. Recognition of these reflexive reasons will also give everyone a reason to act in such a way that allows another to act on such reasons in her own case.

To summarise, an individual person can be the realiser of a property (humanity, beauty, wit, knowledge, moral worth) that is a good in itself. There are, we agree, no reasons, based merely on instantiating such properties, that turn on their first-person character. These properties provide the basis for universal reasons: if some person instantiates the good of humanity, no one person, who recognises the good of humanity, has any more reason to act in recognition of that property than another who similarly recognises the good. Our claim is not that there is some property that is good in itself that only I in fact have—my me-ness, separate from my humanity—and that this gives me an extra reason to be concerned about myself. My total goodness does not reside in the sum of my humanity, plus my me-ness.19 The value that underwrites self-concern is not a form of goodness in that sense. Nevertheless, an individual can have a concern for herself in virtue of being that for whom any action or object has a particular value, instrumental or not. These are reasons—based on my capacity to determine reasons in the setting of ends, and securing of means—that turn on their first-person character. The nature of my being, not as a means, nor as an end, but rather—as we have put it somewhat poetically—as a beginning gives me—we think—non-instrumental reasons that turn on their first-person character.

If we think about the reasons for self-concern as grounded in this way in our capacity as determiners of our action in accord with practical reason, we note that it has certain structural features that will serve to distinguish it from other kinds of concern. On this account (i) self-concern is not due to my being an object of concern: in the context of my determining what it may be good or useful to do there is no choice to be made between being concerned for one person, rather than another, picked out as a particular object of concern because they are the determiner. It is, accordingly, (ii) non-comparative concern: there is no being concerned for one person rather than, or less or more than, another. It is a (iii) distinct kind of concern from concern of another: self-concern understood in this way is the privilege and burden of any rational agent in the context of her own action, and needs to be understood in this way. It is (iv) universal self-concern in the sense that anyone who understands the proper basis of self-concern understands that others also share such reflexive concern. They are also the determiners of what may be good or useful for them to do. It is a concern one recognises, even if one does not share, in another.

3.3. The argument from the self as the site of action

One attractive feature of grounding self-concern in this way on the agent as the primary source, and product, of her determinations about what to do, is that, once it is accepted, it gives us scope to articulate related reasons for well entrenched intuitions of time-concern, and reality-concern. These in turn cast light on a further reason why I should matter to myself, as the locus where my capacity to set ends is exercised.

It seems to us that that how things are now is of special importance, and important in a way that how they were, and are yet to be, isn’t. Going to the meeting now matters differently from having to go to it later, or having gone earlier. We think that the person who fails to see the claim of the present, as a claim of a different and distinct kind from the claim of the past and future is in some way irrational, even though it is sometimes hard to explain to them in what way beyond re-iterating something like ‘this is now, then was then, and that is yet to be’. A NOW-CONCERN claim, parallel to SELF-CONCERN, seems justified.

NOW-CONCERN: The fact that an event is occurring now is a reason for me to want, or not to want, that event. This reason derives from the effects of the event on how things are now, not its effects on anything else. And its force as a reason turns on its temporal character.

Moreover, and even less contentiously, how things actually are seems to be of special importance. Actually going to the meeting matters differently from counterfactually going to it. We think that the person who fails to see the claim of the actual, as a claim of a different and distinct kind from the claim of the counterfactual, is in some way irrational, even though it is sometimes hard to explain to them in what way beyond re-iterating something like ‘this is actually happening, that only might have’. An ACTUAL-CONCERN claim, parallel to SELF-CONCERN, also seems justified.

ACTUAL-CONCERN: The fact that an event actually occurs is a reason for me to want, or not to want, that event. This reason derives from the effects of the event on how things are actually, not its effects on anything else. And its force as a reason turns on its modal character.

Our special concern with the actual, and with the present, is rooted in the fact that agents capable of rational action must set themselves to act in this world, now. Setting aside our role as human agents, it might seem that we have no reason to take any time, or any world, as having a distinct significance in virtue of being this world, now. However, an agent can only themselves act, in the present time, in the actual world. The present time and the actual world are of special importance because they are part of the site of agency: now, here in the actual world, is the only context where I can immediately move.

It is not that me, or now, or the actual have categorical goodness properties that make them special. That is not the source of the reasons for their specialness: the reason for their specialness flows from the role they play in being the site of a subject’s own determinations about actions to perform.20 The fact that something is actually, happening to me, now is a reason for me, as an agent, to care about it because it is actually, happening to me, now.

4. Self-Love, Egoism, Selfishness, and Self-Concern

In this final section we want to do two things. We want to clarify what it means to say that our account gives us SELF-CONCERN. We will do that partly by showing how our understanding of self-concern relates to, and contrasts with, what Setiya calls ‘self-love’.

So, we have argued that we have identified a type of practical reason that is non-instrumental, and whose force turns on its first-person character. However, one could object that this type of reason is not a reason for caring particularly about one’s well-being, or (a fortiori) for thinking that one’s well-being matters more than that of others.

We grant that. But this leads us to the last part of our diagnosis, our point (D). We want to argue that placing such construals—welfarist, egoistic—on SELF-CONCERN is a substantive and questionable move.

What is Setiya’s notion of self-love? Self-love, unlike the special type of concern for oneself that we have argued for, specifically does not turn on my thinking of myself as me:

It is not irrational for me to love myself, whatever I am like, and so to take a disproportionate interest in my own well-being. The justification for doing so is not that I am me, but the fact of our shared humanity. (2015: 469)

However, ‘this justification is “impersonal”’. (2015: 469)

Self-love is the way Setiya tries to do justice to the intuition of self-concern while rejecting the SELF-CONCERN principle. Love, for Setiya, is the source of a non-instrumental, special concern for someone:

a common element of most [forms of love] is disproportionate concern for the interests of the beloved. (2015: 469)

Moreover, the fact that someone is human is enough reason to inspire, although not to require, love. So, self-love is justified, for Setiya, as a trivial consequence of there being a justification for loving everyone. An individual’s self-love is just a particular case of the universal love she has reason to have for all. Thus, an individual’s self-love is a source of concern for that individual, but it is not a source of concern for herself that turns in a non-trivial way on her first-person relation to herself.

However, Setiya’s notion of self-love falls far short of capturing the intuition of self-concern. Whatever one thinks of our account of self-concern, self-love lacks the basic form required.

First, self-love is insensitive to the first/third-person asymmetry in the intuition. All the participants in the trial are human beings, and equally worthy of love. If I am an altruist, I may well love them as much as, or more than, I love myself. So, the move from knowing that one of us will suffer to knowing that I will suffer need not make a difference to how I care. But the Intuition is that it will: realising that the unfortunate participant in the Trial is me necessarily makes a difference to how the outcome matters.

Second, self-love is contingent. I might have reason to love anyone because they are human; but by Setiya’s own lights, this reason is non-insistent: it makes it permissible but not mandatory to love them. As he puts it, ‘the fact of another’s humanity is sufficient to justify love, though not to require it’ (2015: 469). But then this goes for me, too. I wouldn’t be wrong or irrational in loving myself, but I don’t have to, and would not be being irrational in not loving myself. And, indeed, some actually don’t. Think of some deeply depressed persons, exhibiting a lack of self-care and having no love for themselves. If self-love were the only source of self-concern they should lack a concern for what happens to them. But this runs against the modal shape of the Intuition of self-concern. When I come to realise that the unfortunate participant in the trial is me, it necessarily makes a difference to how the outcome can be taken to matter, however riven with self-hatred the participant.

Third, self-love also makes my grasp of another’s self-concern optional. But part of the intuition of self-concern, we think, is that if you learnt that someone else in the trial got unlucky, grasping that they think of themselves as ‘me’ would make you grasp that the news will be of special concern to them. To feel special concern if the bad outcome falls on me goes with understanding that another subject will feel the same concern for themselves if the bad outcome falls on them. This however doesn’t flow from self-love. Almost the opposite does. Love, in Setiya’s own view, is a matter of ‘disproportionate concern for the . . . beloved’ (our emphasis), a matter of allocating one’s attention unequally. Having self-love should make it less likely that we have love for others, and are able to put ourselves in their shoes. In any case, understanding self-love is understanding that it’s not necessary, so finding it in ourselves won’t enable us to expect it in everyone else.

Finally, self-love seems unmotivated. Perhaps trying to mitigate the view’s implication of contingency, Setiya argues that it is particularly natural to love oneself, because this is ‘the primordial case of love at first sight.’ He amplifies thus:

Self-love is the primordial case of love at first sight. Or better, since I am available to myself not just perceptually but through immediate knowledge, in both agency and introspection, it is love at first act, or first thought. I am presented to myself in a special and primitive way in which I am presented to no-one else: as the agent of my actions and the thinker of my thoughts. What could be more natural than to love the person who is given to me in that way? (2015: 469)

But this particular justification for self-love is in some tension Setiya’s own argument that there is no reason to care for the object of IMMEDIATE KNOWLEDGE. If it makes sense to wonder ‘Why care so much about the one you know first-hand?’ (2015: 467), then is it just as legitimate to ask, ‘Why love the one you know first-hand?’

We think the problem is this: Setiya starts from a broad intuition of self-concern, but doesn’t discuss how it should be interpreted, or in detail what general principle of self-concern we should draw from the cases eliciting the intuition. His thinking on self-concern involves some background decisions that are not theoretically innocent; it also gets pulled in various directions during the course of the discussion.

The starting point is a question about whether there are reasons for self-concern that are essentially first-personal reasons: ‘the central feature of self-concern [is] that it is essentially first-personal.’ It is about ‘the special status, for me, of the person I am’ (2015: 455) The second point at issue is whether, if there are first-personal reasons for self-concern, they reflect a non-instrumental value we place in ourselves: ‘reasons of self-concern: the non-instrumental interest we have in what benefits and harms us in the future’ (2015: 448).

Setiya takes this to present us with the following key question: ‘Do facts about what will happen to me provide me with reasons for preference whose force turns on their first-person character?’ (2015: 462).

There is a reading of preference here, which Setiya seems happy to accept, that implies more than reasons for self-concern of a special sort, but also comparative reasons: reasons for ranking. They are reasons for preferring oneself to others; for putting oneself first: ‘According to self-concern, first-person thought plays an essential role in the justification of disproportionate interest in oneself’ (2015: 469, our italics).

Finally, Setiya stipulates that the reasons in play must be welfarist reasons. They are reasons for preferring well-being over harm; for wanting what happens to me to be good for my well-being, rather than bad for my well-being: ‘reasons of self-concern [are reasons] whose normative explanation stops with my well-being’ (2015: 445).

We think that in a discussion of self-concern these four features of reasons—first-personal, non-instrumental, comparative, welfarist—should not be taken to stand or fall together.

We aim to disentangle these features, using some careful distinctions laid out in Richard Wollheim’s Thread of Life. Wollheim’s discussion of self-concern is extremely helpful because it takes the trouble to contrast self-concern with the closest things it is not. Of self-concern Wollheim says:

Self-concern does not involve egoism. Self-concern involves the thought that, for instance, my future states are important to me in a way that yours aren’t or couldn’t be: but not the thought that my future states are more important to me, let alone the thought that they are more important, than yours are or could be. Furthermore, self-concern implies the thought that your future states are important to you in the very way in which my future states are important to me. (1984: 243)

Wollheim’s statement of self-concern captures the central feature of the intuition of self-concern—its essentially first-personal dimension—but emphasises those properties that we found self-love lacked—the difference in the kind of concern involved in self-concern, in concern for another, and in the recognition of self-concern in others.

Wollheim’s setting aside of ‘egoism’ is precisely the setting aside of the comparative component that crept into Setiya’s characterisation:

The attitude that egoism requires of a person towards his desires, or towards those of his [future] states which directly depend upon them, is to believe that they are more important . . . than the desires of others, or the states dependent upon them. (1984: 242–43)

To have self-concern, namely to be concerned about my future because it is mine, does not involve egoism:

To think self-concern involves egoism is on a par with thinking that, because I insure my property and not yours against fire, this shows that, if there were a fire I should prefer your property to burn down to mine doing so. (1984: 244)

Wollheim also seeks to distinguish self-concern from ‘selfishness’.

A person’s desires are selfish insofar as they are directed towards his well-being, his good name, his mental or bodily health as such . . . and the person is selfish insofar as he is motivated by such desires. (1984: 243).

This is the welfarist feature that Setiya reads into his SELF-CONCERN. But, again, it isn’t entailed by self-concern as an essentially first-personal type of concern.21 A resistance fighter is concerned about her future as hers, but this might mean being concerned about doing the right thing, or doing what she ardently feels compelled to do, even if this takes sacrificing her well-being, her good name, and indeed her life. There are other values (beyond well-being) that self-concern is indexed on—fairness, truth, dignity, for example.

So, the first lesson we can draw from Wollheim is that the various ways Setiya elaborates his SELF-CONCERN are just different notions, that don’t flow from a single source. In particular, the central first component, that self-concern is concern for myself as me, does not entail either the third or the fourth components, the egoist/comparative and the selfish/welfarist elements that Setiya’s stipulation has the effect of adding on without defence. Having special concern for oneself, as oneself, doesn’t entail that one thinks of oneself as more important, or of one’s well-being as what matters especially.

A second observation. The comparative and welfarist components Setiya reads into SELF-CONCERN are just those he takes to be involved in self-love: ‘a disproportionate interest in my own well-being’ (2015: 469, our emphasis). But they are just the wrong components to figure in a general principle of self-concern. A welfarist and comparative notion, as we’ve seen with self-love, fails to explain the intuition of self-concern.

So, the second lesson is that it’s the other two ingredients which should go into a principle of self-concern. What the intuition of self-concern supports is the thought that what will happen to me matters in a special way—matters differently—and perhaps also that it matters full stop, not conditional on some further reason or other. But the intuition does not support the thought that it matters more, or matters only insofar as it might affect my well-being in particular.

Of course, if I, as most do, also have egoistic and selfish desires—and my well-being matters to me more than that of others, as it matters to most of us, most of the time—then my learning that my well-being will be adversely, or advantageously affected, will, if I am self-concerned, also come to be that which I am concerned about when I am concerned about what happens to me. But these are not features of self-concern as such.22

We think it will clarify things if we rephrase the principle of self-concern slightly. Here is Setiya’s initial definition again:

SELF-CONCERN: The fact that an event will benefit or harm me is a reason for me to want, or not to want, that event to happen. This reason derives from the effects of the event on my well-being, not its effects on anything else. And its force as a reason turns on its first-person character.

We do not object with the letter of that principle, but we think it invites confusions and that Setiya does, in fact, read into it some elements that it doesn’t entail.

What should be changed? The first sentence should be read non-distributively: ‘The fact that an event will [benefit or harm] me is a reason for me [to want, or not to want], that event to happen.’ What matters is that the event will happen to me—and in so doing, it will probably either benefit or harm me. But what isn’t prejudged is whether my self-concern will dictate a preference for one or the other: sometimes self-concern turns on values like truth or justice or dignity that don’t necessarily pull in the same direction as my comfort.

The second sentence misleadingly suggests a welfarist (and perhaps comparative) interpretation, but Setiya makes clear just after the definition that what he tries to capture here is the non-instrumental dimension of reasons of self-concern: ‘the reasons in question do not derive from the instrumental significance of my well-being for some further end’ (2015: 445).

So, to make the meaning of self-concern more conspicuous, we propose the following reformulation:

SELF-CONCERN*: The fact that an event will affect my future states is a reason for me to care that it will happen, a reason that what will happen to you isn’t or couldn’t be. It is a non-instrumental kind of reason, that stops at the fact that it will happen to me. And its force as a reason turns on its first-person character.

This principle is just what our discussion of the source of our self-concern, as determiners of our actions, invites: it captures an essentially first-personal and non-instrumental concern for oneself. And it succeeds in accounting for the Intuition of self-concern just where self-love failed: because self-concern so understood flows from being the determiner of my actions, it isn’t contingent; and it makes non-optional my grasp of self-concern in another agent who I know also thinks of herself as herself determining what she is to do.

5. Conclusion

Does the way I think of myself as me make a difference to how I care about what happens to the person that is me? Do we have HARMONY* at last? We agree with Setiya that the way we think about ourselves and the grounds for reasons for self-concern should be calibrated. However, we think that this shows us two things—one about first-person thought and one about self-concern—that Setiya does not.

One, it shows us that the relations we bear to ourselves in first-person thought are not just epistemic. We also relate to ourselves as agents, and a recognition of that is needed to make sense of self-concern. Moreover, the way in which we relate to ourselves as agents shows us that we do not, cannot, in the basic case take an instrumental attitude to our value. Our value is pre-suppositional on there being anything to value.

Two, it shows us that there are reasons of self-concern: how things are with me has a special claim for consideration in my practical deliberations, and the relevant type of consideration is not instrumental. But these reasons are not reasons in themselves to care especially for my well-being, or to put myself first.

Reasons of self-concern, in sum, are pre-suppositional reasons that turn on their first person character. But they are not, thereby, egoistic or selfish reasons.


A Marie Skłodowska Curie Post-Doctoral Research Fellowship (Phenoself, FP7-PEOPLE-2013-IEF project n°622127) enabled us to work together at UCL; this paper is the result of a talk prepared for the concluding conference. We are extremely grateful to the funders, and to UCL for being such a supportive and effective host. The paper has benefitted greatly from the excellent comments we received from two referees for this journal, and we received very helpful editorial guidance from Nick Jones. We were helped enormously by feedback from a number of people. We are especially in debt to Tyler Burge, Jeremy Fix, Conor McHugh, Daniel Morgan, and Peter Sullivan, whose comments resulted in substantial changes. We also learnt much from the audiences at various conferences, including the ‘Mineness’ conference at Logos in Barcelona; the KBNS ‘Knowledge and the a Priori’ conference in Stirling; the ‘Concept of Self’ concluding conference of Phenoself at the Institute of Philosophy, London; and the UCLA Colloquium. Related ideas building on this work were presented at the SIFA conference in Novara, at the conference on ‘The Nature of Conscious Subjects’ in Freiburg, at the Manchester Philosophy Seminar, and at the ‘Perspectives on the Minimal Self’ conference in Vienna, where again audiences offered very helpful comments.


    *The authors are equal co-authors listed alphabetically.

  1. Note that we have identified the harm likely in this case as one of bodily damage, and not in terms of pain. It might be thought that there is an explanation of why one will be concerned about one’s own pain in a different way from the way one is concerned about another’s that comes directly from the nature of pain: only one’s own pains are painful; others are distressing in some further way. This explanation does not obviously turn on specificities of first-person thinking; sensations of pain may exert their motivational force without the need for the mediation of thought. But our specific object, and Setiya’s, is the practical significance (if any) of first-personal thought. Our aim is not to provide a general account of the grounds of self-concern, but only to ask whether the first-personal way of thinking about oneself provides such a ground. We leave aside here the question whether there is any interesting articulation between self-concern based on first-personal thinking, and other sources of self-concern, and how our account of self-concern relates to other potential ones.
  2. We follow Setiya (2015), with the caveats in Section 4, in understanding non-instrumental reasons of self-concern as those reasons ‘whose normative explanation stops with’ how things are with me—instead of turning on the ‘instrumental significance’ of my predicament ‘for some further end: the general happiness, say, or the fulfillment of my obligations’ (2015: 445).
  3. We elaborate on Setiya’s notion of self-love in Section 4.
  4. An illustration: a thought including a spatial indexical, like ‘here is where my glass is’, is based on a relation of physical contiguity with the object. This relation is what gives the thought its rational significance, explaining why it makes sense to transition from having the thought to reaching for the glass. (Adapted from Setiya 2015: 466.)
  5. Additionally, I am related to that individual’s experiences in a way quite unlike the way I am related to others’ experiences: I may know about the latter, but the former are the only ones I have. Experiential relations may also be reflected in the first-personal way of thinking about oneself (see Guillot 2016) and may provide a further argument for self-concern, that we leave for further work.
  6. Setiya makes room for a subject’s agential relation to herself, but only in so far as it falls under her knowledge of herself, which includes knowledge of her intentional actions. But we think agency merits separate consideration. However, even if it turned out that all there is to my relationship to myself as agent is knowledge of my own intentional actions, we think that would still offer a way to resist Setiya’s conclusion. In the interest of space, we won’t explore here whether the principle of IMMEDIATE EFFECT, below, implies some kind of agential knowledge falling under IK (we think it does—see O’Brien 2007); or whether IMMEDIATE EFFECT is indeed merely a specification of IK (we think it isn’t). Either case would be a starting point for pursuing our strategy (C) (see p. 6) in countering Setiya’s ‘why care’ argument.
  7. This is the point of Evans’s insistence in the Varieties of Reference, that ‘“I”-thoughts are thoughts in which a subject of thought and action is thinking about himself—i.e., about a subject of thought and action. It is true that I manifest self-conscious thought, like “here”-thought, in action; but I manifest it, not in knowing which object to act upon, but in acting. (I do not move myself; I myself move)’ (1982: 207).
  8. One might want to say that in the typing example, there is a sense in which both hands, the unimpaired as well as the impaired one, are instruments of writing words. And one might even want to say the agent could view her unimpaired hand as good or useful for writing with, in a way that contrasts with the way her impaired hand, or indeed her leg was good or useful for typing with. But that is not for the subject to meet the objectual constraint: what I don’t and cannot do is assess whether to use myself, my limbed body, as the best means to act with towards my chosen ends, over some other means. My embodied self, here, shows up as the subject of action rather than as an object mediating my action. We thank an anonymous referee for inviting us to clarify this point.
  9. We thank Jeremy Fix for helping us reformulate the constraint.
  10. This is not to deny that one could have an instrumental attitude towards a target that is not an object, but (for example) a fact. I have an instrumental desire for the state of affairs that my friend be given anaesthetic before his operation tomorrow. (We thank an anonymous reviewer for suggesting this example, and for raising this issue.) Our focus here is not on instrumental concern generally, but on instrumental concern for objects, because this is the shape of our concern for ourselves as instruments.
  11. This is not to deny that awareness of what I am able to do, or good at doing (including by comparison with others) can lie in the background of my deliberations about what to do. But thinking about what lies within my means is still not thinking about myself as a means or as an object. Where my abilities are implicitly factored in, this bears on my decision whether to φ at all; it doesn’t turn the question into one about whether to φ with myself or with someone else. Seeing a reason for someone else to do what needs doing (perhaps because they are better suited to the task) is not seeing a reason for me to act with or via them; neither, then, is seeing a reason for me to act the same as seeing a reason to act with myself. Thanks to Jeremy Fix for raising this point.
  12. This point is made by Matt Boyle (2016) and is reflected in the distinction there between ‘subjective’ and ‘objective’ first-person attitudes. Ninan (2016) gives a more general argument for thinking that the attitudes involved in agency don’t, by default, have the agent as an object, but represent only ‘agent-neutral’ actions (Section 3.3).
  13. Cf. Kant (1785/1998: 165–87), Korsgaard (1996: 122), Wood (1999: 124–32), Sussman (2003).
  14. It is disputed by Regan (2002) and Theunissen (2018), among others. We thank an anonymous referee for pressing us on this.
  15. There is even room for it (in a non-rationalist reading) in a utilitarian view like Peter Singer’s, where the moral consideration we (and other animals) deserve is due to our having interests, i.e., a capacity for valuing (1975/2015: 7–8).
  16. We thank Conor McHugh for pressing us on this.
  17. Considerations of non-objectuality also imply a further contrast between self-concern so understood, and concern for another rational agent A. The former is implicit in non-objectual deliberations about φ-ing, whereas the latter typically arises from objectual appreciations of what is at stake in A φ-ing. See feature (i) below.
  18. We thank Conor McHugh for helping us articulate this point.
  19. See Caspar Hare (2009) for considerations to the contrary. Here we are using the term ‘me-ness’ in a metaphysical sense, distinct from the phenomenological sense sometimes given to the term in debates about consciousness, as described for example in Guillot (2017). We discuss in ongoing work the separate question whether phenomenal ‘me-ness’, if it exists, could also play a role in self-concern.
  20. This line of reasoning also explains why we care about the future. In agents who have the capacity for mental time-travel, now-concern implies future-concern. Being able to predict that events still ahead of us will, in the future, be ‘now’, and that we will care about them in a special way, gives them a particular weight in our present plans.
  21. Setiya explicitly chooses to focus on self-concern expressed as concern for my own well-being, so our claim here isn’t that he is relying on unacknowledged assumptions; it may be that Setiya would see some of our points as broadly compatible with his argument. Still, we believe there is scope for substantive debate in how self-concern is initially framed. Our point is that his choice to restrict the focus of the investigation, so that only narrowly welfarist expressions of self-concern are taken into account, is open to criticism. We think that it is arbitrary (self-interest is just one of a range of expressions of self-concern, and probably the least mysterious) and that it unduly preempts the shape of the discussion by ruling out at the outset a possible source of self-concern that doesn’t turn on categorical goodness properties, but hinges instead on the structure of agency. We also note that Setiya provides little initial guidance on how we should think of ‘well-being’, or of the related ‘benefits and harms’ that may be reasons to want it; it is only at the end of his paper that it is made fully clear that for Setiya, as perhaps for Wollheim, reasons for preferring well-being are understood as disjoint from those that flow, say, from ‘personal projects’, ‘relationships with others’ or indeed concerns with dignity or integrity. While this takes us beyond the scope of this paper, we are in fact highly sceptical that concerns for my well-being can be separated from such other concerns in any such clear-cut way.
  22. For the avoidance of doubt, our point is not to deny that there are distinctively first-personal reasons to be concerned about one’s wellbeing—far from it. Indeed, the rationality (pace Setiya) of concern for my own wellbeing, turning on the fact that it is mine, is an implication of our account, to be explored more fully in further work. Self-concern as we define it is an agent’s special responsiveness to her own non-instrumental value as a source of value. A creature which has intrinsic value is surely one whose good is to be desired, for its own sake; and the good of creatures like us certainly includes wellbeing on any definition (see previous footnote). Our view, rather, is that self-concern encompasses far more than concern for our wellbeing; that the inclusion of wellbeing under the scope of self-concern is a contingent fact about the kind of agents we are, rather than an essential feature of self-concern; and that an exclusive focus on wellbeing puts us on the wrong track by holding the discussion captive to a limited expression of agency—the pursuit of self-interest—and diverting it from the subject’s place in the structure of intentional action more generally.


1 Boyle, Matt (2016). Knowledge as Subject and Agential Knowledge. Unpublished manuscript.

2 Burge, Tyler (2000). Reason and the First Person. In Crispin Wright, Barry C. Smith, and Cynthia Macdonald (Eds.), Knowing Our Own Minds (243–70). Oxford University Press.

3 Campbell, John J. (1994). Past, Space and Self. MIT Press.

4 Dickie, Imogen (2015). Perception and Demonstratives. In Mohan Matthen (Ed.) The Oxford Handbook of Philosophy of Perception (p–p). Oxford University Press.

6 Evans, Gareth (1982). Varieties of Reference. Clarendon Press.

7 Guillot, Marie (2016). Thinking of Oneself as the Thinker: The Concept of Self and the Phenomenology of Intellection. Philosophical Explorations, 19(2), 138–60.

8 Guillot, Marie (2017). I Me Mine: On a Confusion Concerning Subjective Character. Review of Philosophy and Psychology, 8(1), 23–53.

9 Hare, Caspar (2009). On Myself, and Other, Less Important, Subjects. Princeton University Press.

10 Heck Richard (2002). Do Demonstratives Have Senses? Philosophers’ Imprint 2. Reprinted in The Philosopher’s Annual 25 (2002).

11 Kant, Immanuel (1998). Groundwork of the Metaphysics of Morals. Edited by Paul Guyer. Rowman & Littlefield. (Original work published 1785)

12 Korsgaard, Christine (1996). The Sources of Normativity. Edited by O. O’Neill. Cambridge University Press.

13 Ninan, Dilip (2016). What Is the Problem of De Se Attitudes? In M. García-Carpintero and S. Torre (Eds.), About Oneself: De Se Attitudes and Communication (86–120). Oxford University Press.

14 O’Brien, Lucy (2007). Self-Knowing Agents. Oxford University Press.

15 Regan, Donald H. (2002). The Value of Rational Nature. Ethics, 112, 267–91.

16 Salje, Lea (2016). Knowing Selves (Doctoral dissertation, University College London).

17 Setiya, Kieran (2015). Selfish Reasons. Ergo, 2(18), 445–72.

18 Singer, Peter (2015). Animal Liberation: A New Ethics for Our Treatment of Animals. The Bodley Head, Vintage, Penguin Random House. (Original work published 1975)

19 Sussman, David (2003). The Authority of Humanity. Ethics, 113(2), 350–66.

20 Theunissen, Nandi (2018). Must We Be Just Plain Good? On Regress Arguments for the Value of Humanity. Ethics, 128(2), 346–72.

21 Wollheim, Richard (1984). Thread of Life. Cambridge University Press.

22 Wood, Allen (1999). Kant’s Ethical Thought. Cambridge University Press.