Agency

By all accounts an agent is a system that acts to bring about certain goals or purposes. These are states of affairs that it generates on a reliable basis, and not just by accident: on the basis, for example, of having been selected or designed to realize the goals or to follow certain procedures that identify goals to realize. But agency requires more than the pursuit of purposes. Otherwise the sunflower plant that tracks the sun – that reliably moves so as to maintain its orientation towards the sun – would count, implausibly, as an agent.

As we think of agency in our ordinary exchanges, the most obvious requirement over and beyond the pursuit of certain goals is that this pursuit should be maintained over variations in the circumstances of the system: variations that are more significant, intuitively, than those in the angle of the sun that the sunflower has to cope with. A system may count as an agent and be so simple that it has only a single goal. But still, it must be able to pursue that goal across different scenarios, adjusting its behavior so as to realize the goal under the particularities of each situation.

Take the simple robot that is constructed to pursue the goal of raising certain objects on a flat surface to an upright position (List and Pettit Reference List and Pettit2011, Ch 1). Such a robot will have to be equipped with some apparatus for determining whether any glass or cup or bottle on the surface – say, a tabletop – is on its side or upright; think of this as an eye-like receptor that continually scans the objects on the surface. And then, presented with a bottle on its side, the robot will have to be organized across variations in the size and shape of the object, and its distance and direction, to adjust behaviorally so as, things going well, to put it upright: this, presumably, with the help of wheels for moving, levers for lifting, mechanisms for grasping, and so on.

In order for the robot to adjust its behavior appropriately in different scenarios –and count thereby, we may suppose, as an agent – it must be able in some sense to register the particularities of each scenario. It must change in response to receptor inputs, and its changed state must then serve to shape the behavior – and to shape the behavior as it evolves in time (Hurley, Reference Hurley1998) – so as to achieve the goal. This is to say, in other terms, that the robot must form representations of how things are in each scenario where it acts, and let those representations direct its behavior there.

The general lesson, in familiar Humean terms, is that an agent is a system that reliably acts to fulfill its goal-seeking states according to representations that it reliably forms: in perhaps a deflationary sense of those terms, it acts so as to satisfy its desires according to its beliefs. The agent's reliability in forming beliefs constitutes epistemic rationality, its reliability in acting so as to satisfy desires constitutes practical rationality. The system may not be unfailingly reliable or rational but, if it is to count properly as an agent, then it must generally fail only when circumstances are abnormal by independent criteria. Thus, the robot may be misled under certain lighting about the position of an object; and it may not manage to put an object at the edge of the table upright: it may knock it to the floor. In the first case, it would fail epistemically, in the second practically.

This simple image of agency is not uncontroversial but it is at least familiar, being supported with variations by recent philosophers like David Lewis (Reference Lewis1983), Robert Stalnaker (Reference Stalnaker1984), and Daniel Dennett (Reference Dennett1987). Others may place demands on agency over and beyond the requirements it encodes: they may demand capacities, for example, associated here with personhood. But still, it represents a model of agency that is an intelligible reconstruction of our commonsense assumptions, if not the only reconstruction admissible (Pettit Reference Pettit1993, Ch 1).

However simple, of course, our model of agency allows of many specifications, applying in different ways across the spectrum from robot to animal to human. Where we human beings pursue purposes of highly distinctive, contrasting kinds, for example, other agents may act only in pursuit of a few simple goals. Where we employ highly sophisticated representations of different types of situation, and represent how things are, were or will be, as well as how they may or must be, other agents may not range much beyond the here and now. And, of course, where we human beings use a common language to express and shape our purposes and representations, robots and animals do not do so in the same way.

Selfhood

Every agent, by virtue of what agency involves, treats itself as special. It acts on the basis of the memories and beliefs that its own experiences support, not the experiences of any other agent. It forms desires on the basis of the sorts of things that attract it – its likes and dislikes – and not those of any other agent. And it acts on the basis of the intentions it itself forms, not the intentions that materialize in any other. Talk of experience, memory and attraction may be out of place with the robot, suggesting that it must be conscious. But even the robot treats itself as special in forming beliefs on the basis of the inputs it registers, in determining situation-specific goals on the basis of its general, hardwired goal, and in acting to pursue those situation-specific goals that it settles on.

It is noteworthy that every agent must treat itself as special, not just at a time, but over time. This entails that whether X at time t-n is the same agent as Y at the later time t shows up in the fact that X's experiences, tastes and intentions shape in a unique way the memories, beliefs, intentions and actions of Y. And whether Z at time t + n is the same agent as Y at the earlier time t shows up in the fact that similar relations to those between X and Y hold between Y and Z.

In order to treat itself as special in such ways, every agent, even one as simple as our robot, has to form representations or beliefs about itself. Thus, if the robot acts intentionally so as to put a glass on the table into an upright position, it will have to move towards that glass and at a certain point form a representation, whatever its exact content, to the effect that the glass is within reach; at that point, all going well, it will stop moving and reach out for the glass. The responses to this representation that the agent displays in ceasing to move and in reaching for the glass will be programmed into the robot, of course, but if we are to make sense of them from within the intentional stance – if we are to see them as the responses of an agent – then we must characterize the representation in a way that makes them rational. That means that it must be a representation or belief to the effect, from its standpoint, that the glass is within my reach, that it is graspable by me. The belief must register a relationship between the glass and the robot, not a property of the glass alone, and if it is to be expressed in words, as from a first-person viewpoint, the formula used has to deploy the first-person indexical ‘I’ or a variant (Perry, Reference Perry1979).Footnote ¹

As even the simple robot will have to form this sort of belief about itself, so it will also have to form a range of other beliefs, if it is to act as it is designed to do. It will have to form beliefs to the effect that this or that object on the table is out of reach (of me), that it is to (my) right or left, that it is of the right size (for me) to lift, and so on. Does the fact that every agent has to form such beliefs about itself mean that it has a self in any sense? Surely not. Although it forms beliefs about itself, this simple sort of robot is little more than an impersonal mechanism, not something to which we could ever plausibly ascribe a self: it is a handy tool, not a colleague.

The beliefs that the robot must form about itself fail to argue for its having a self, plausibly, because they do not identify the self on which they bear. A belief will identify something for an agent when it picks it out from among other potential entities as a subject of which it predicates this or that property or relation. Thus, viewing it from within the intentional stance, we must credit the robot with identifying this or that glass as a subject of predication and then assigning a property or relation to it when it forms the belief that the glass is on its side.

The robot does not identify itself as a subject of predication in this manner, despite believing that it has certain relations to the objects on the table: that this or that glass is near to it or graspable by it, or whatever. The robot does not pick itself out from among other agents present or possible, and identifying itself among those candidates, form various beliefs about its relationships to things on the table. In positing those relationships, the robot's self-beliefs, as we may call them, lock onto itself; they do not first pick it out and then register that it, the subject identified, bears those relationships. The self-beliefs lock on to it as the only agent they could possibly apply to, and they apply to it of necessity.Footnote ²

The robot's subject-predicate belief that a certain glass is on its side may err in either of two ways. Among the glasses on the table, it may misidentify the particular glass that constitutes the subject of the belief that it is on its side; a pulsating light may lead it to attribute that property to an upright glass nearby. Or, assuming that it has no problem in identifying the targeted subject, it may misattribute the property of being on its side; the glass may appear to be on its side due to its unusual balloon shape. Because its self-beliefs do not first identify itself and then attribute to it a certain relationship with one or another object on the table, they cannot err on the first count. They may misattribute that relationship, taking an object to be nearer than it really is, for example. But they cannot misidentify the agent to whom the relationship is ascribed: in a case like this, as Gareth Evans (Reference Evans1982) puts it, there is no error through misidentification.

These observations invite us, by way of contrast, to consider an agent that is able to form identifying as well as locked-on beliefs about itself. An example would be an upgraded robot that is able, not just to form self-beliefs of the kind illustrated, but also to form beliefs that other entities are robots like itself, and in particular to form beliefs to the effect that it, the entity to which it has a locked-on, identification-independent relationship, is one of those robots. It can form beliefs that might warrant expression in first-person formulae like the following: I am this robot here, not that one there; I am the one currently moving towards the glass on the right, not the robot moving towards the glass on the left; I am the one who will reach its destination first; and so on.

We could reasonably say that this robot exists for itself in a way in which the earlier version did not. It connects in the same locked-on way with itself but identifies the agent with which it connects in that way as one among many possible candidates, and it treats the target of those two sorts of beliefs as one and the same. It exists for itself not just as the unmissable target of an indexical thought but also as this or that fallibly identified robot. It exists for itself, not just as a private reference point, in other words, but also as an entity located among other entities in a public world and characterized, like those entities, by various properties or relations.

It is plausible to suppose that an agent with a self, as distinct from an agent without a self, is at its most basic, an agent that exists for itself in this way: a pour-soi, in Sartre's (Reference Sartre1958) terminology, not an en-soi. It is an agent that not only locates things in the public world, as presumably any agent must do, but that also identifies itself in that public world. Employing the private mode of self-reference available to any agent in the formation of beliefs and other attitudes, it identifies that agent – it identifies itself, so referred to – with a figure in the public world, characterized by public properties. It thinks thoughts of the form: that is me, that is how I am!

As we have described this upgraded robot, the beliefs it forms about itself are sophisticated in two ways. They are stimulus-independent beliefs insofar as they include beliefs beyond the here and now, as in the robot's believing that it will get to its destination sooner than others (Camp, Reference Camp2009). And they are compositional beliefs insofar it is true that any property the robot ascribes to itself, it can ascribe to others, and any property it ascribes to others it can apply to itself (Evans, Reference Evans1982, 104). Might there be agents who can form a belief like ‘That is me’, but only in a non-compositional, stimulus-dependent way? The question bears on the self-recognitional capacities of those non-human animals who identify themselves in mirror reflections but we may put the issue aside in the present context (Boyle, Reference Boyle2018).

Personhood

Selfhood is associated with persons in particular and it's important that we introduce this category as well as that of agency. There is widespread agreement that not every agent is a person but little or no agreement on what it is that makes some agents into persons, others not. Some may hold by the traditional view, deriving from Boethius, that persons are agents of a reasoning character or nature – rationabilis naturae – or agents with similar, general capacities like the ability to communicate, to interpret others, to be aware of themselves in relation to others, and to enter reciprocal arrangements (Dennett, Reference Dennett1979).Footnote ³

Many identify persons, however, not by their general capacities, but by specific, normatively relevant features. Thus, Locke (Reference Locke1975, s26) takes it that persons can assume and incur responsibilities, arguing that ‘person’ is ‘a Forensick term appropriating Actions and their Merit’. John Rawls (Reference Rawls2001, 23) argues that what distinguishes persons is that, of necessity, they can bear and assert rights: they are ‘self-authenticating sources of moral claims’. And Harry Frankfurt (Reference Frankfurt1971) identifies persons – as distinct from ‘wantons’ – as agents that engage the autonomy ideal of acting on desires that they desire to be moved by.

There is an approach to the characterization of persons that helps to explain the appeal of these different accounts and has a claim to our allegiance on that ground. In any case, it is the account we shall work with here. Deriving from some brisk remarks by Thomas Hobbes (Reference Hobbes and Curley1994, Ch 16), it takes persons to be agents who have and exercise the capacity to ‘personate’ in relation to one another. Agents personate insofar as they assume a special authority in communicating their attitudes to others and, at least in general, live up to their words in practice: they act as the attitudes communicated would warrant. They speak for themselves, inviting others to trust their words: to rely on their displaying the profile or persona that they project in what they say (Pettit, Reference Pettit2008).Footnote ⁴

The personation account entails the presence of the capacities that Boethius and Dennett and others ascribe to persons: agents can hardly speak for themselves in the manner envisaged without such abilities. But the account also explains the normative features that others take to be distinctive of persons. Connecting with Locke, it would make sense of why persons, inviting the reliance of others, must assume and incur responsibilities to prove reliable. Connecting with Rawls, it would explain why each person must have certain rights: the right to personate with others, for example, as well as the right to rely on others to live up to their personation. And, connecting with Frankfurt, it would make sense of why personhood implies an ideal, if not quite that which he envisages; here the associated ideal is that of displaying in actions the persona projected in words.Footnote ⁵

What is it that makes people's communication of their attitudes into a form of personation? Why are persons said to assume authority for the communication, inviting others to rely on them? The key to the answer lies in the notion of commitment.Footnote ⁶

We can define someone's being committed to holding by a certain attitude or to performing a certain action as their communicating that they have that attitude or that they will perform that action in a way that exposes them to a higher expense than normal – and so makes the communication more credible than normal – should they fail to act on the advertised disposition. Let the standard of normality be set by the expense they would incur if they merely reported on their disposition, as they might report on the disposition of a third person. Any communication of a current attitude or a projected action will be commissive insofar as it involves voluntarily incurring a higher risk of loss in the case of a miscommunication.

Suppose, unusually, that I were merely to report a belief or desire. Suppose that I said in a suitably tentative manner that I think I hold such and such a belief or desire or that I think I will perform such and such an action. Then, as with any report, I would be able to offer one of two salient face-saving explanations – one of two excuses – for failing to act accordingly. I might claim that I was misled by introspective evidence about my attitudes: my mind misled me. Or I might maintain that I changed my disposition since reporting it: I changed my mind.Footnote ⁷

There are two prominent forms of commitment, avowals and pledges, that correspond to these excuses, and they figure prominently, as we shall see, in personation. The avowal of an attitude manifestly forecloses appeal to the misleading-mind excuse, thereby making the avowal more expensive and credible than a report. The pledge to perform an action – alternatively, the pledging of the intention to perform it – manifestly forecloses appeal to either the misleading-mind or the changed-mind excuse, making it more expensive and credible than even an avowal (Pettit, Reference Pettit2018a).

I will avow a belief that p, communicating the presence of that belief-state, if I just assert that p on the basis of having made up my mind, presumably in light of the evidence or data, that it is the case that p. The avowal will communicate, not just that p, but also that I believe that p, insofar as it is a matter of shared assumption that, absent insincerity or incompetence, anything I assert I also believe. But I do not learn that I have the avowed belief thereby communicated by introspecting the contents of my mind and relying on the evidence of what I find there; I do so by relying on the capacity manifested in making the assertion that p: the capacity to make up my mind. And so, the avowal manifestly precludes me from excusing a failure to act as if I had the belief by invoking the misleading-mind excuse. Not having been led by evidence about my mind, after all, I couldn't have been misled by it. Thus, the avowal of the belief will communicate the belief more expensively and more credibly than a detached report would have done.

As reporting on what I take to be the case will amount to avowing the belief that that is the case, so reporting on the presence of manifest desiderata in a certain option, reporting that it is the fairest or the most exciting alternative, for example, will generally constitute an avowal of that desire. I will form or reinforce the desire in fastening on the operative desiderata, letting them elicit or entrench the desire, as I will form or reinforce a belief in attending to the data that support it. By citing the desiderata, moreover, I will generally convey that I desire that option – why would I have cited its appealing feature otherwise? And I will convey this in a way that leaves me without access to the excuse, should I act as if I did not have the desire, that I must have been misled about its presence.Footnote ⁸ Not having been led by introspective evidence that it was present – knowing that it is present by virtue of knowing what I did in fastening on the desiderata – I cannot have been misled by such evidence. Thus, the communication will count as commissive in the same way as the avowal of a belief.Footnote ⁹

I may avow an intention to do something, say to go to a concert with you, on the same desiderative basis as with a desire: ‘that would be great fun’, I may say, in response to a query about joining you at the concert. But in the case of an intention, and of the corresponding action, I may do something more. Recognizing that you won't go to the concert unless I join you, for example, and expecting to want to join you, I may make a pledge to be there. In doing so I will manifestly foreclose the changed-mind excuse as well as the misleading-mind excuse for having misled you, should I not turn up. You will naturally say ‘But you promised!’ if I try to excuse not turning up by saying that I changed my mind.Footnote ¹⁰

Personation in broadly the Hobbesian sense involves commitments of the kind that avowal and pledging exemplify. When persons speak for their attitudes, they assume the authority that goes with their purported ability to communicate those attitudes while foreclosing the possibility of excusing a miscommunication by appeal to a misleading-mind mind or, where appropriate, to a changed-mind. To the extent that they put aside face-saving excuses for failure, persons will stake their reputation on living up to their words; they will bet on themselves to prove reliable in that way.

Commitments of those kinds are common in social life, figuring prominently in conversational exchanges, whether about what is or might be the case or about what the participants should individually or jointly do. But commitments are even more common than this suggests, because commitments can assume a virtual as well as an active form. People will make commitments virtually when they fail to say ‘Nay’ to the manifest expectations of others; they will make them actively when they say ‘Yea’ in order to put novel expectations in place.

Suppose you and I live in a society, for example, where it is manifest to all that certain regularities hold and that everyone is expected in general to live up to them without question: this, for example, in avoiding violence, deception, fraudulence, infidelity and the like. I will know in the range of everyday interactions that others manifestly expect me, without any question, to accept those norms and to be willing to conform; indeed, I will also know that, should I fail, then appealing to excuses of the misleading- or changed-mind sort will not wash. This being a manifest matter between us, my not rejecting those expectations will communicate that I acquiesce. And by acquiescing in those expectations, I will effectively avow acceptance of the norms and pledge conformity to them. Indeed, the same will be true across the spectrum of manifest, unrejected expectations that I and others form in dealing with one another.

Thus, by the account adopted here, persons are agents that personate in their relations with one another, and do so to the point where personation, active or virtual, is an inescapable aspect of their individual lives. As a byproduct of commissively communicating their dispositions – not, as an effort in narcissistic self-portrayal (Strawson, Reference Strawson2005) – they will each shape a persona or image of themselves that they project and generally seek to honor. They will hold out this persona to others, as if in proclaiming: this is who and how I am; this is who and how I back myself to be.

Personation is a social activity, by this account, and personhood can only materialize among a group of agents, not in the solitary individual.Footnote ¹¹ Hobbes (Reference Hobbes and Curley1994, 26.6) would seem to endorse this view, holding that it is impossible to make a commitment to yourself on the grounds that ‘he that can bind, can release’. Even if we admit that the practice of commitment presupposes social life, however, induction in that practice will presumably enable persons to commit to themselves in an analogous, if not exactly similar way: to form resolutions, as we say. When I sincerely make commitments to others, indeed, there is a sense in which I will commit to myself at the same time; I will internalize the persona I project; I will think as well as say, ‘this I who and how I am’. I may commit and personate insincerely in many social contexts, of course, and at a barely imaginable limit, I might even do this in all. I may make commitments that I do not internalize, in other words, impersonating a self that I do not have.Footnote ¹²

By our earlier argument, agents may achieve self-identification and selfhood, and achieve it in a sophisticated fashion that is compositional and stimulus-independent, without being persons in the sense of personating agents. The self-identification of persons is bound to be similarly sophisticated: they will each be able to predicate various properties of themselves, whether they see themselves as ‘I’ or under a public identifier; they will be able to predicate those very same properties of others, when that is appropriate; and of course they will be able to do this in abstraction from current stimuli.

But apart from being sophisticated in that way, the self-identification of persons is going to be special in at least three respects. Self-identification in the personal case will characterize the self substantively, projecting a full attitudinal and practical profile. It will do this inescapably, as an essential part of what it is to personate. And it will do it as a matter of aspiration, not accomplished fact: it will hold out an idealized version of the figure with which the person identifies, albeit a version that they commit to realizing.

The referenced self

Assuming that I am able to identify myself in public terms, and to do so in a sophisticated way – in a compositional, stimulus-independent manner – the most basic profile that I will be able to assign to myself under that aspect is as the agent referenced in my intentional states. This is the agent whose experiences give me memories and beliefs, whose likes and dislikes lie at the origin of my desires, and whose intentions dispose me to perform corresponding actions. This agent is one among many subjects who populate the world and, if I ask myself who I am, then that agent provides one obvious answer to my question: this is who and how I am; this is me.

What is it about this referenced agent, to pose a metaphysical question, that makes it me? If I pose the question in the present about an agent in the past or indeed an agent in the future – if I ask about myself qua agent and not, for example, qua organism – then there is only one answer possible. This or that agent in the past or future is me myself now, just insofar as there are the familiar agential linkages between the three. The claim is not that I have those linkages with the past or present entity because they are me. Rather, they count as me – we count as the self-same individual over time – because those linkages obtain. What I am over time is constituted by the chained time-slices that connect in that way.

The chaining of those particular time-slices will be explained, no doubt, by all sorts of facts about my neural make-up; if my brain now were linked electronically with different brains in the past or the future, after all, then the path of the chain would be quite different. But there is an important sense, to use Derek Parfit's (Reference Parfit1984) expression, in which there need be no deep fact about what gives me my self-identity – my identity as this referenced self – over time. Plausibly, the connections between the different links make it the case that I am present in each link and that, over time, I am constituted by the chain they form. There would be a deep fact about who I am, and about what binds my temporal stages together, only if things were the other way around: only if it was my presence at each linked point – my presence in some independent mysterious sense of me – that explained why the chain ran through just those locations.

If the metaphysics takes this form, and there is no deep fact about what makes me the referenced self I am, then various well-known science-fiction possibilities are open. I might survive as the same self in this sense, if I were tele-transported, for example, having my body reconstituted in duplicate form. And equally I, as I am now, might divide in the future, with distinct selves at the later time sharing a chain up to the moment of division, and with each referencing the links they have in common as they form relevant memories or act on relevant intentions. But I as a referenced self could hardly survive fusion with another, since that future agent would have conflicting connections into the past: it would have to serve the masters provided by rival experiences, tastes, and intentions. And equally, in this particular identity, I could not survive the loss of agency that a vegetative state would imply.Footnote ¹³

Moving to an epistemological question, how can I know myself as an enduring agent in this self or profile? As an enduring agent, I will be characterized, not by the states I happen to be in at any moment, but by the chain of connections with the states of past temporal stages and, presumptively, later stages in the future. Thus, there won't be much about this enduring self that will be revealed as I introspect at any moment. Things will be more or less as Hume (Reference Hume1978, I.6.3) famously describes them: ‘when I enter most intimately into what I call myself, I always stumble on some particular perception or other, of heat or cold, light or shade, love or hatred, pain or pleasure. I never can catch myself at any time without a perception, and never can observe anything but the perception’.Footnote ¹⁴

But while I may not have much to learn about the nature of this referenced self in any form of direct introspection, there is still a sense in which it is bound to be accessible to me. This is the self to which I orientate, after all, when I try to remember a past experience, when I deliberate about what at some future point to do, and when I feel anxiety at the prospect of a visit to the dentist.Footnote ¹⁵ It contrasts with the agent to which I orientate when acting as a member of a group, for example, seeking to realize group goals according to group assumptions (Pettit, Reference Pettit2018b). It is a self I will see out of the corner of my eye, so to speak: in apperception, as we might call it, rather than perception. I know it, not by acquaintance and not by description, but as a self, referenced in all my agential adjustments. It is revealed to me in the way in which the viewpoint from which a landscape photograph was taken is revealed by a picture, despite that fact that it does not figure in the photograph itself.

Turning now to the practical question, how far does the referenced self matter to me, or how far ought it to matter? In the nature of the case, I am bound to care about getting the experiences of this self right as I try to remember something, about keeping this self on an effective path in pursuing my goals, and about its not suffering too much in the visit to the dentist. But the observation needs to be qualified in a way that parallels a remark made in discussing the metaphysical question.

I do not care about my referenced self in this way because it is me in an independent sense. Rather, this self counts as me because it is the target of such care. If we can speak of self-love here, it is the innocuous form of self-love that Rousseau describes as l'amour de soi, as distinct from l'amour propre; it is a form of concern with self that he takes to be part and parcel of our nature (Dent, Reference Dent1988).

Derek Parfit (Reference Parfit1984) suggests that as I look further and further into the future in tracking the agent with whom I am linked in the fashion described, there will be fewer and fewer linkages between me now and that agent. He concludes that it may make little sense, then, for me to be moved prudentially by the interests of that agent: say, if I am young, by their interest in having a good superannuation to rely on. This does not follow with the sort of self-concern appropriate to the referenced self. Any connections of the relevant kind will mean that that future agent is me, even if those connections all go through intermediate stages: even if they do not include, as they surely may include, a direct connection like that established by an intention now that I enjoy a good superannuation then. And insofar as that future agent is me, however distant in time, it will attract my corresponding concern as a matter of necessity, not as a function of an optional – and, as it might seem, questionable – degree of prudential concern.

Imagine a future, then, in which I cease to exist properly as an agent. I enter a vegetative state and, while I retain my identity as an organism, I cease to be an agent and cease, a fortiori, to be the same agent as I am now. On the approach taken here, I cannot care in the agential way for that individual in the future. I can only care for that human being, as I might care for someone else, in a relatively altruistic and, as we might say in this case, a prudential manner.

The personated self

The personation that makes me into a person, by the account offered earlier, is an activity that I pursue as an agent, identified to myself in the first-person manner just sketched. But that activity itself consists in a form of self-identification and directs us to a distinct profile in which I may recognize myself: it yields a different self, as we may say, that I, one and the same agent, may bear. This different self is that which I assume in my second-person relationships with others, and that which I self-ascribe when I am being sincere with my audience.

When I identify myself in personating or personal terms, I do not do so in the spirit of a self-reporter or autobiographer. I do not stand back from myself and describe for one or another audience – perhaps just for myself – the sort of figure I cut in the interpersonal world where I rely on others and invite them to rely on me. Such an autobiographical pen picture would carry no special authority. Indeed, given the bias or partiality that we each feel for ourselves, it might be especially suspect.

In writing as a self-reporter in this way, I could claim only the epistemic authority of someone determined to seek out all sources of evidence and to be responsive to the evidence mustered. In personation, however, I assume a distinct practical form of authority, claiming the capacity to commit only to attitudes that I can enact and to enact all those attitudes to which I commit. I stake my reputation for displaying this commissive-enactive capacity when I avow any beliefs, desires or intentions, putting aside the possibility of invoking a misleading-mind excuse for a miscommunication. And I do so in a yet more demanding way when I pledge an intention to act in one or another fashion, foreclosing appeal to either a misleading-mind or a changed-mind excuse for having been misleading.

In the case of either sort of commitment, of course, I may occasionally fail to enact the attitude to which I commit and there may even be un-foreclosed excuses that I can offer to save my reputation. I will be able to excuse a failure to act on an avowed belief or desire, or indeed an avowed intention, by a change of mind. And I will be able to excuse both a failure to act on an avowed attitude and a failure to execute a pledged intention by certain unforeseeable changes of circumstances: bereavement, accident, illness or whatever. When I have no excuse to offer for a failure, however, the only alternative will be an apology and, in token of sincerity, a renewal of the commitment. And if the failures become too frequent, I will jeopardize any claim to have the commissive-enactive capacity associated with personhood; I will begin to look like what Frankfurt calls a ‘wanton’ rather than a person.Footnote ¹⁶

The profile of myself that I present here constitutes a personated self that is distinct from the referenced self. Assuming sincerity and competence, this is a self that I will actually prove to bear. And assuming sincerity and competence, it will have to be a relatively unified self, not one committed in inconsistent ways. I may personate on different fronts with different audiences but I cannot personate sincerely – I cannot personate for myself, so to speak – in diverging, unreconcilable ways.

The personated self is metaphysically unproblematic. It is a character or persona that I create for myself in the exercise of commissive-enactive competence. It is grounded, then, in two aspects of my performance in relation to others – and to myself considered as if I were another. First, in the things I say, or the things I let go as not needing to be said, in speaking for myself. And second, in the things I do in giving life to the character to which my speech testifies. Where the referenced self is guaranteed to exist by the exigencies of agency, the personated self is something I construct to begin with and reconstruct when occasion demands.

To return to a point made earlier, however, this self is not a construct that I intend to create as such. Some theorists hold that human beings actively construct a narrative of their lives and of who they are, and that this is an inevitable aspect of personhood. That claim ties personhood, implausibly, to a highly intellectualized form of reflection and a pattern of self-scripting that sounds downright narcissistic, as critics have suggested (Strawson, Reference Strawson2005). But the idea here is not open to such objections, as the character that each of us is said to construct for others, and indeed for ourselves, is a byproduct –generally, we may presume, an unnoticed byproduct – of an independently intelligible practice of commitment.

That I construct and reconstruct my personated self in this manner does not preclude my deconstructing it, of course, as when I disown certain attitudes that I previously avowed or some intentions and actions that I previously pledged. Given changes in the data or desiderata I access, or in my appreciation of them, it is entirely intelligible why I should change beliefs or desires I previously avowed. And given unforeseen circumstances, or transformative experiences (Paul, Reference Paul2014), it is equally intelligible why I should change pledged intentions or policies as well.

Turning from the metaphysical to the epistemological issue, how do I know the personated self I assume and display in the manner described? The answer, in brief, is: by means of a maker's, not an observer's, knowledge. In exercising my commissive-enactive authority, I make it the case that I commit only to attitudes I can enact and that I enact all the attitudes to which I commit. Insofar as I know what I am doing in such an exercise I will be in a position to know the personated self I project, although I may not be conscious of it as such.

The exercise of my commissive-enactive competence may take the form of active avowals or pledges or of acquiescence in the manifest expectations of others. But in either case the knowledge of what I am doing will give me a base for knowing who in the personal sense I am; it will be capable of revealing my personated character to me. The problem of how I know what I am doing in any activity is a recognized problem, of course, but assuming it is soluble in the general case, it will yield a solution to the specific epistemological issue raised by my personated self.

But while my personated self will be knowable to me for these reasons, it may take effort to achieve a full knowledge of who and how in this sense I am. That will require a unified sense of the different fronts on which I am committed and of the package that those commitments constitute. Thus, it may take time and trouble for me to develop such a sense of where I am committed. There is going to be a point, therefore, to the exhortation in this case to know yourself; I may have to practice a degree of discipline and meditation to achieve self-knowledge in this sense.

Finally, to the practical issue. Ought I to care about myself in this personal guise? And am I likely in any case to do so? The answer in each case is affirmative.

I cannot help but care about living up to the commitments I make sincerely to others, and the self I thereby project. My reputation as someone others can rely on, after all – indeed my reputation as someone I can myself rely on – depends on my displaying such fidelity. For me to make commitments sincerely without any care for whether I proved reliable would be impossible; it would undermine the very notion of commitment.

But not only would it be unlikely that I should not have any care for keeping my commitments, and remaining faithful to my personated self. Such a lack of care would also be undesirable both for me and for others. It would make it impossible for others to be able to rely on me, and impossible for me to be able to elicit their reliability in return. It would make for a loss on all sides.

The case for being faithful or true to the personated self over time is memorably put by Shakespeare, of course, in words he ascribes to Polonius, when bidding farewell to Laertes, his son. Polonius may be presented in an otherwise unflattering light by Shakespeare but he is surely credited with wisdom, when he says to Laertes:

This is wisdom, at any rate, if Polonius has the personated self in view. To be true to the personated self is just to live up to active and virtual commitments, letting your words and actions testify to the same persona. And, as a result of that convergence, it is to speak reliably to others, avoiding falsehood or duplicity.

One final comment, however, on the personated self. The fact that I unify this self, know it properly and care for it appropriately does not mean that it is morally admirable. For all that is required by the account offered, I might assume and enact the identity of a Nietzschean Übermensch, who treats certain others with disdain and makes serious commitments only to an elite of perceived equals. The ideal of the personated self is a structural ideal, not an ideal of a substantive kind. It teaches a lesson about how I should be – unified, self-knowing and stable – regardless of the character that I actually have.

The imputed self

In personation I am invested in establishing a persona in the minds of others that, when I personate sincerely, I can identify with: I can see as me. But for each of us there is a character that is not sculpted in personation, or that is sculpted only partially in that manner. This is the image of me that exists in the minds of others as they view me at a distance, so to speak, beyond the reach or control of my second-person avowals and pledges. It is a character created by third-person gossip about my attitudes and dispositions, by the available record of my achievements and failures, by the labels and stereotypes under which I am seen in our society and, if I am exposed to publicity in any domain or degree, by what becomes accepted as a matter of common or public assumption about me.

This character, alien and uncontrolled, may not seem like a candidate for being seen as me. But this is the self, known under my public name, that those with whom I do not regularly interact will take to be me. And since the way those others treat me will reflect the character that they ascribe, I cannot detach myself wholly from this identity; I cannot pretend it is someone else. Whatever the self imputed to me, of course, I may take different attitudes towards it: I may relish it, I may rail at it, or I may regard it with relative indifference. But even when I am indifferent, I cannot ignore it completely; it follows me as a shadow in social space, whether I like it or not.

The inescapability of this imputed self is nicely captured by Jorge Luis Borges (Reference Borges1962, 246–47) in a short essay, entitled ‘Borges and I’. Seeing himself cast as a result of his publicity in a persona that is created by others as much as by himself, he comments: ‘The other one, the one called Borges, is the one things happen to’. While this Borges is beyond his control, it is not a figure he resents. ‘It would be an exaggeration’, he reports, ‘to say that ours is a hostile relationship; I live, let myself go on living, so that Borges may contrive his literature, and this literature justifies me’.

Where Borges displays an amused indifference to his imputed self, the protagonist in Jean Paul Sartre's (Reference Sartre1948) essay ‘Portrait of the Anti-semite’ represents someone who relishes it. This young man, unsure of his identity, is taken to be hostile to Jews, perhaps quite mistakenly, and finds that the image impresses acquaintances and prompts them to make allowances for his attitudes. Enjoying that recognition, then, he comes to cherish the image in which he finds himself cast. He will do this, Sartre suggests, not necessarily because the image appeals independently, but because it makes him into a somebody: it gives him bearings by which to navigate and rescues him from a sea of existentialist indecision. The lesson is a general one. Each of us is an unconstrained center of decision-making, not someone predestined or predetermined to have certain attitudes or to act in certain ways. And as a result, we are each subject to the temptation to espouse in bad faith any character we find imputed to us.

But the self imputed to us, whether we like it or not, may also be one we resent and reject. If I belong to a disadvantaged group, for example, and am cast in a stereotype of a religious, ethnic or gendered kind, then I am extremely likely to resist that aspect of the character foisted upon me. And equally, if I am defamed – if, at the limit, a negative characterization is imposed on me, as a matter of common awareness in my community – then I must shrink from the character imputed.

In either case, but especially, in the stereotyping one, the image pinned on me may not only run counter to how I personate sincerely and reliably; it may silence any attempts to correct the image imposed and to change the expectations raised (Langton, Reference Langton2009). It may deprive me of the capacity to speak for myself in a way that commands the credence of others, discounting or recasting the sorts of things I might say. It may compound the injustice of the misrepresentation, in other words, with the epistemic injustice of disabling me from putting it right (Fricker, Reference Fricker2007).

The fact that others invariably give us a character means that there is a third sort of self in which I or anyone else may be invested. This imputed self does not have the same interest as the referenced or personated self but it is still worth putting on the map. While the metaphysical and epistemological questions it raises are not very significant, there are real issues about why it matters, and about how far it ought to matter, to us.

Metaphysically, the imputed self is constituted by the opinion in which others hold me, independently of how I personate, and will appear in as many fractured and unfriendly forms as such opinion assumes. Insofar as I sincerely personate, and want my imputed self to conform to its personated counterpart, this will give me an interest in unifying the imputed self and bringing it into line with how I personate. But there may be strict limits to how far I can hope to achieve that result, creating a single image on this front of who and how I am.

Epistemologically, the imputed self is bound to be elusive, being constituted by the independent opinions that people hold of me. I may be painfully aware of a prejudicial category in which I am cast, or of some personally defaming gossip. Or if I am a public figure, I may take pleasure in the positive things that are said about me. But I will generally find it hard to gauge the opinion in which I am held by others. While people may gossip among themselves about me, after all, they are unlikely to gossip to me about myself. If someone is a friend, they may tell me what others are saying about me, whether to congratulate me about the good that is said or to warn me about the bad. But gossip proper belongs essentially to the third-person perspective on my performance rather than to the second-person relationship in which I can personate with my interactants.

Metaphysically fractured and epistemologically elusive though it is, however, my imputed self is bound to matter to me. And with good reason. Like others, I am likely to care about how people think of me, this being part of our social nature. But even if I don't worry about their opinion of me as such, I will certainly care about how that opinion leads them to treat me. Thus, I have good instrumental reason for caring about the self imputed, when there is any danger of being individually defamed or of being cast with others under a prejudicial label. And equally, I have good instrumental reason to care, if I occupy a public position – say, a political office – and success in my sphere of activity, electoral or otherwise, depends on how I am viewed beyond the bounds where I may hope to personate.

Assuming that we can put such special cases aside, however unrealistic that assumption may be, is there good reason in general to care about my imputed self? The question is not whether there is reason, to welcome the imputation of a positive character or to bemoan the imputation of a negative; clearly, the answer there is, ‘yes’. The question rather is whether there is reason, first, to desire to shape the imputed, un-personated self and, second, to try to shape it by filtering or massaging the information available about me – by running a personal public-relations exercise. By long and solid tradition, the answer to both questions is, no. The standard wisdom is that there are reasons against trying to shape this self, beyond the bounds where personation is possible, and reasons against even wanting to make such an attempt.

The reasons against trying to shape the imputed self in a personal public-relations exercise derive, at base, from the fact that this will put me in zero-sum competition with others. Whatever the domain of performance, I will want to establish how well I do in comparison with others, not on some absolute scale; after all, it is the performance of people generally that will set the standard in any domain (Brennan and Pettit, Reference Brennan and Pettit2004). This means that trying to shape my imputed image may put me in fruitless competition with others. We may each make costly efforts to improve our relative standing when those efforts may cancel out one another, leaving us individually no better off than we were before we started.

The reasons against even desiring or wanting to make such efforts – the reasons against caring, in that sense, about the imputed self – are many. One is that indulging that concern may distract me from focusing on something intuitively much more important: the self that I forge for others and for myself in personation. A second is that it may be impossible ever to tell if that concern is satisfied: the epistemic elusiveness of the imputed self means that trying to see if it is in an appealing shape may be as fruitless as trying to paint a picture of someone in the dark. And a third consideration is that if the concern is to be satisfied, I had better keep it hidden from others; no one is going to achieve a high standing in the opinion of their peers if, beyond the context where defamation or prejudice threatens, they are seen to be concerned to achieve such standing. ‘The general axiom in this domain’, as Jon Elster (Reference Elster1983, 66) says, ‘is that nothing is so unimpressive as behaviour designed to impress’.Footnote ¹⁷

A final reason against even wanting to try to shape the imputed self is that it is an inherently insatiable concern: it can drive me to seek, not just to do well by local standards, but to outscore all others, and by as much as possible. As Rousseau (Reference Rousseau2020, 232) argued, following Hobbes, ‘the first feeling excited by this comparison is the desire to be first’. It is this love of being first, this love of pre-eminence, that Rousseau castigated as l'amour propre, as distinct from the innocent l'amour de soi. Seeing such self-love as a quest for supremacy, Kant (Reference Kant2006, 167) also condemned it, finding it present in ‘the manias for honor, dominance, and possession’: in these manias, he said, ‘the human being becomes the dupe’.