Equal Opportunity, Responsibility, and Personal Identity

  • Ian Carter


According to the ‘starting-gate’ interpretation of equality of opportunity, individuals who enjoy equal starts can legitimately become unequal to the extent that their differences derive from choices for which they can be held responsible. There can be no coercive transfers of resources in favour of individuals who disregarded their own futures, and no limits on the right of an individual to distribute resources intrapersonally. This paper assesses two ways in which advocates of equality of opportunity might depart from the starting-gate interpretation. The first involves limiting the degree to which people are liable to pay the costs of their past choices. The second involves limiting their initial opportunities so as to prevent certain risky or apparently short-sighted choices. The paper compares these alternatives in terms of their compatibility with a particular conception of persons as morally equal and temporally extended. It constructs this conception by combining reductionist premises about personal identity with the premise that our status as equals is based on the fundamental requirement of opacity respect. Two conclusions about equality of opportunity are shown to follow from this conception of persons as morally equal and temporally extended: the first is that an individual’s liability to pay the costs of her past choices does not diminish over time; the second is that the individual’s initial scope of choice, in bringing about intrapersonal distributions between her current and future selves, can nevertheless be permissibly limited. The two conclusions are consistent, and the second allows for departures from starting-gate equality of opportunity.


Equality of opportunity Basic equality Responsibility Personal identity Agency Freedom 

1 Introduction

Advocates of equality of opportunity typically recognize the justice of unequal outcomes. On the face of it, unequal outcomes are unproblematic from the point of view of egalitarian justice as long as those involved enjoy a level playing field and the subsequent differences in their situations can be put down to choices for which they can reasonably be held responsible (Arneson 1989, Dworkin 2002, ch. 2). However, this intuition comes under pressure when the stakes are high and when the period between opportunities and outcomes is long. To cite a familiar, if extreme, example: suppose a group of contemporaries received a very substantial endowment 50 years ago when they were aged 21, that this endowment was shared equally among them as a lump sum in lieu of most of the benefits normally accorded by a more traditional welfare-state, and that one of them squandered her entire share shortly afterwards in a casino. Suppose, further, that this lone gambler satisfied all the normal requirements for being held responsible for her choice: she possessed certain basic agential capacities, she was not coerced into gambling, and she was perfectly aware at the time of all the possible consequences and their probabilities. She is now a destitute 71-year-old and, let us further suppose, the risky choice she made at 21 is the principal cause of her current plight. Is her situation morally problematic from the point of view of egalitarian justice?

If you believe in equality of opportunity and yet also feel that there is something unjust about the plight of the destitute 71-year-old, you might try qualifying equality of opportunity in one of two ways (Fleurbaey 2005, Bou-Habib 2006, Stemplowska 2011, Tomlin 2013). First, you might argue against the view that people with equal initial opportunities should always be liable to pay the full costs of risky or apparently short-sighted choices made in the past. This alternative involves an ex post correction of the unequal outcomes that follow equality of opportunity, and with it a limitation of the ‘responsibility sensitivity’ of egalitarianism. Second, you might argue that people’s initial opportunities, though equal, need not be so extensive as to include the freedom to make such risky or apparently short-sighted choices. For example, an initial opportunity set might include a compulsory insurance package. This second alternative involves an ex ante restriction of people’s opportunities, but does not limit their ex post liability to bear the costs of the choices they do make.

My aim in this article is to assess these two ways of qualifying the ideal of equality of opportunity. I shall criticize the first – that is, limiting ‘responsibility sensitivity’ – but defend the coherence of the second – that is, limiting initial opportunities. The challenge I set myself is to arrive at this conclusion while nevertheless starting from certain foundational liberal premises that seem, on the face of it, to point in the opposite direction: that individual persons have an equal status as agents capable of making choices and planning their own lives, that such planning might well involve the intrapersonal distribution of resources between different stages of one’s life, and that respecting people as agents with this capacity to plan involves according them freedom.

My argument will rest on two major building blocks, and much of this article will be devoted to their exposition. The first building block consists in an account of the basis of equality – of what it is that makes us equals, such that we have a basic moral right to equality of opportunity (Carter 2011a). The second building block consists in an account of personal identity over time – one that rejects the standard view of persons as perfectly unified agents extended over whole biological lives. I shall try to show that these two building blocks can be combined in such a way as to support my favoured interpretation of equality of opportunity.

I shall take as my initial point of reference the limiting case of a version of equal opportunity that completely rules out both of the above-mentioned limitations – both the ex post limitation of liability for past choices and the ex ante limitation of initial opportunities. I shall call this version “starting-gate equality of opportunity”. According to this view, egalitarian justice requires an equal playing field at the start of people’s lives (on the simplest model we focus on a single generation of exact contemporaries), after which individuals may legitimately become unequal as a result of their choices and no one is bailed out except through the voluntary choices of others. Starting-gate equality of opportunity has been defended in various forms by so-called ‘luck-egalitarians’ and ‘left-libertarians’.1 In policy terms, it is normally interpreted as implying a preference for ‘basic capital’ or ‘stakeholding’ in comparison both to basic income and to more traditional egalitarian state benefits.

As the extensive literature testifies, while some advocates of equality of opportunity have taken a strong stance either for or against the starting-gate view, many others have found themselves torn. Members of the latter group appreciate the emphasis of the starting-gate view on respect for people’s agency and choices, but they shy away from some of the implications it is normally taken to have in terms of liability to pay the costs of past choices. They would like to justify placing some normative limits on those implications, but they are not quite sure how to do so in a coherent way. My arguments are addressed to these ambivalent egalitarians. I assume, from the start, the validity of a principle of equal opportunity grounded in equal respect for persons understood as beings capable of making choices and planning their lives. The question I address is whether this assumption can be made compatible with policies that limit either people’s available choices (ex ante) or their liability to pay the costs of their actual choices (ex post). There are of course egalitarian bases for imposing such limits that involve either rejecting a respect-based principle of equal opportunity or supplementing it, in a pluralist spirit, with rival welfarist assumptions. Such alternative bases steer clear of the challenge I have set myself and therefore lie outside the scope of this article.

My argument proceeds as follows. In Section 2, I present an account of equal moral status which, I believe, serves to ground the principle of equal opportunity. This constitutes the first of the two building blocks mentioned above. In Section 3, I expand on the egalitarian quandary described above and show how it implies a dilemma regarding the issue of personal identity over time. In Section 4, I attempt to find a way out of that dilemma by integrating the theory of basic equality set out in Section 2 with a particular account of personal identity. This account of personal identity constitutes the second of the two building blocks mentioned above. In the fifth and final section, I show how these building blocks can ground a principled solution to the egalitarian quandary concerning the roles of choice and responsibility in equality of opportunity, arguing against the ex post limitation of responsibility for past choices but in favour of the ex ante limitation of initial opportunities. In this final section I contrast my own conclusion with that of Patrick Tomlin (2013), who in a recent article in this journal has argued in favour of the ex post limitation of responsibility for certain past choices, starting from premises about personal identity that have much in common with my own.

2 Grounding Equal Opportunity in our Basic Equality

The standard assumption of liberal, freedom-based theories of justice is that individuals are entitled to freedom, or opportunity, in virtue of their possession of certain basic agential capacities.2 Such basic capacities include the cognitive capacities to recognize valuable ends, to exercise judgement in selecting among competing ends, and to select the best means to given ends; and volitional capacities to follow through in pursuit of chosen ends (strength of will, the capacity to control emotions, and so on). Unfortunately for the egalitarian, none of these capacities appears to be possessed equally among human agents. Our capacities for discrimination, emotional management, and so on, vary in degree and are possessed unequally. Why, then, ought opportunity to be accorded equally to persons in possession of these capacities, rather than unequally, in proportion to the degrees of such capacities? What, in other words, is the source of our ‘basic equality’ – of the equal status that grounds our entitlements to distributive equality – if all the potentially relevant properties, excluding those with a religious or fantastical basis, appear to be scalar and possessed unequally? If we are unable to point to an equally possessed, morally relevant property, our endorsement of equality of opportunity will seem to be groundless.

A first step in solving this problem is to say, with John Rawls (1971, §77), that each person has the property of having the above-mentioned capacities at a level that lies within a certain range – that is, in the range from a minimum level upwards. This ‘range property’ – of having at least a minimum of the relevant capacities – is binary and is therefore possessed equally by all those who do possess it, given that the relevant capacities are either inside the range or outside it. Rawls calls this range property ‘moral personality’. Possession of moral personality qualifies one to be considered an equal and therefore as a party to the social contract on the basis of which egalitarian principles are selected.

There remains, however, the task of justifying the moral relevance of the range property of moral personality. What is so special about falling within a range? Why focus on this particular property, thus favouring the treatment of individuals as equals, rather than focusing directly on the scalar properties upon which Rawlsian ‘moral personality’ supervenes, thus favouring the treatment of individuals as unequals?

Elsewhere I have argued that our basic equality depends on a kind of respect, called ‘opacity respect’ (Carter 2011a). To adopt the attitude of opacity respect toward a person is to recognize that she possesses certain basic agential capacities while also refraining from taking into account the degree to which she possesses them above a minimum qualifying threshold, where the meeting of that threshold can be safely assumed on the basis of the normal outward behaviour of humans. Respect for a person can be understood as involving the maintenance of a certain distance, adopting an external perspective, not ‘looking inside’ them, and in this sense treating them as opaque. To recognize an absolute minimum threshold while also abstaining from taking into account levels above that threshold is to ascribe salience to a range property. The notion of ‘opacity respect’ therefore provides an independent moral reason for focusing on the Rawlsian range property of moral personality.

Why should the attitude of respect for persons be thought to involve treating persons as opaque in the above sense? The answer, I have suggested, lies in a particular kind of dignity that is protected or conferred through such treatment. There is a sense in which people enjoy dignity when they are considered immune to evaluations of certain kinds. The kind of dignity I am referring to here is not dignity in the standard Kantian sense. There is another kind of dignity that can be both removed and conferred without removing or conferring Kantian dignity – the kind of dignity that depends on others’ assuming our integrity (in the sense of wholeness or completeness), not measuring our capacities or dismantling them into so many reactions to stimuli. This alternative notion of dignity can be called ‘outward dignity’. We confer outward dignity on persons by taking their agency as given and adopting an external perspective. This perspective on agency might not be appropriate in all kinds of human relation, but there is at least a strong liberal case for seeing it as an appropriate attitude for agents of the state to adopt when relating to citizens in their role as bearers of basic rights (Carter 2011a, pp. 556–68).

Not only does the idea of opacity respect provide a plausible grounding for our basic equality; it also coheres with the choice of equal opportunity as an egalitarian principle. If there is a special reason for treating persons as opaque, so justifying our focus on a range property, it is plausible to hold that this reason must be found in the notion of respect. Because respect is an appropriate response to agency, it is plausible to conclude that non-instrumental egalitarianism accords people whatever they are due as agents. At least a part of what is due to persons considered as agents is the freedom to form and carry out plans. This, in brief, is the line of reasoning that leads from respect-for-persons to basic equality, and from basic equality to the principle of equal opportunity.

Some rival accounts of what justice requires us to equalize seem to deprive us of reasons for focusing on a range property, and therefore appear groundless. Consider the case of ‘equality of welfare’. Since welfarism is based on concern or compassion rather than on respect, there seems to be no welfarist reason for focusing on a relevant range property (for example, the property of ‘falling within a particular range of the variable capacity for pleasure or pain’), rather than on the scalar properties on which that range property supervenes (in the example just given, the capacity for pleasure or pain). This helps to explain why Peter Singer (1993), a utilitarian, denies the basic equality of humans. ‘Equality of welfare’ does indeed seem to be a groundless distributive principle.

The above account of basic equality already has some implications for the ways in which the notion of responsibility can be employed when interpreting and applying the principle of equality of opportunity. Any account of equality of opportunity must assume a certain standard of responsibility, where ‘responsibility’ means ‘liability to carry the costs (or enjoy the benefits) of past choices’, and the determination of this standard will depend on how we understand and apply certain other responsibility-concepts. As Nicole Vincent has pointed out in her helpful taxonomy of responsibility concepts, judgements of ‘liability-responsibility’ depend not only on whether we attribute causal responsibility to a person for having brought about an event, but also on the ‘role-responsibilities’ of that person prior to bringing about the event – that is, on what can be reasonably expected of her in moral and prudential terms (Vincent 2011; see also Olsaretti 2009). They depend, for example, on what a person can be reasonably expected to know, or to be able to know, at the time of the relevant choice. And these role-responsibilities in turn depend, in part, on the ‘capacity-responsibility’ we ascribe to the person – that is, on the fact of the person counting as ‘responsible’ in the sense of being in possession of certain basic agential capacities. On the account of basic equality supplied above, this ‘capacity-responsibility’ must be treated as a binary property, because it exists in virtue of possession of the Rawlsian range property of moral personality, the moral relevance of which depends on opacity respect.3 Within the context of the kind of responsibility-sensitive egalitarian theory under consideration, then, role-responsibilities depend only on what can be expected of a ‘normal’ agent – an agent characterized simply as one whose capacities lie within the normal range. Similarly, judgements of causal responsibility will take into account only those facts about an agent that can be permissibly taken into account in the context of relations calling for opacity respect. In particular, they must omit to take into account individuals’ particular degrees, above the threshold, of those basic agential capacities upon which the range property of moral personality has been assumed to supervene (for example, the degrees to which they possess certain basic capacities for rational thought, or the degrees to which they show strength of will), for to assess such capacities above the threshold would be to violate the requirement of opacity respect, and would therefore negate the basic equality that grounds the prescription of equality of opportunity in the first place (Carter 2011a, pp. 567–68).

The notion of opacity respect will prove relevant when we come to address the apparent conflict of intuitions about the merits of starting-gate equality of opportunity, for we shall then need to consider how the above account of basic equality can be extended diachronically. First, however, we should clarify the way in which that conflict brings to the surface the question of personal identity.

3 Equality of Opportunity and Personal Identity: a Dilemma

How should assumptions about personal identity affect our attitude toward starting-gate equality of opportunity? In answering this question, those egalitarians who would depart from the starting-gate view seem to face a dilemma. On the one hand they may assume, as people normally do, that the destitute 71-year-old is the same person as the gambling 21-year-old. In this case, they seem forced to endorse the ‘harsh’ judgement implied by the starting-gate view. Alternatively, they may respond more generously to the plight of the 71-year-old. However, if they believe in equal opportunity because they believe in respect for persons as agents, it seems that the only way for them to justify this more generous response is by denying the identity of the 71-year-old and the 21-year-old. To see more clearly why this is so, let us consider the two horns of the dilemma in turn.

We arrive at the first horn of the dilemma by assuming the common view of a single person as a temporally extended being spanning a whole biological life. On this view, the choices a person makes at the ages of 21 and 71 are choices of the same agent, and the proper object of respect is the agent considered as a single, temporally extended unit. Respect is owed to one unified agent, not to several temporally located bundles of agential capacities considered separately. To view the proper object(s) of respect in the latter way would be to ignore the fact of temporally extended agency: actions take up time; complex actions take up long periods of time; the earlier decisions of a single person can condition in coherent ways the later actions she performs and the later opportunities she enjoys. Persons have the capacity to make long-term plans, including ‘life plans’. To respect them as agents is to respect this capacity and the ways in which it is exercised.

Consider a comparison between the temporal and spatial dimensions of agency. If I own a large plot of land, leaving aside externalities, your respect for my agential capacities surely involves allowing me to decide for myself how to distribute my moveable resources over this land – for example, building one large house in a corner of the plot or two smaller ones a certain distance apart. Similarly, if I am a single, unified agent taking up the time-span of a normal biological life, your respect for my agential capacities surely involves allowing me to decide for myself how to distribute my resources over that time-span – for example, enjoying more opportunities at the beginning of my life and fewer later on (taking risks or living for the moment), or making sacrifices earlier on in order to safeguard later opportunities (investing or buying insurance). Moreover, there is only one moment at which a single agent can make decisions regarding the whole of the temporally extended life she has to lead, and that is at the beginning of her whole life.4 This line of reasoning creates the first horn of the dilemma for those egalitarians who are unhappy with starting-gate equality of opportunity: the standard conception of personal identity, in conjunction with a respect-based account of egalitarian justice, really does entail the starting-gate view.

The temptation, in response to this implication, is to point to the asymmetrical relation between earlier and later selves: my analogy between the spatial and temporal dimensions of agency seems to break down once we see how, under starting-gate equality of opportunity, earlier selves can influence the fate of later selves whereas the converse is not true. However, this asymmetry has moral relevance only if we assume that a person’s later selves have claims of their own that the earlier selves have no right to ignore or to override in their own interests. And to make this assumption is to throw doubt on the commonly accepted view of personal identity. What the commonly accepted view classified as a question of intrapersonal distribution for a single unified agent gets reclassified, on this assumption, as a question of interpersonal distribution between competing rights-holders. Thus, while the assumption that later selves have claims against former selves does indeed ground a rejection of starting-gate equality of opportunity, the way in which it does so is by contradicting the claim that the various selves located at different moments in a single biological life are simply parts of the same temporally unified whole that we call a person – the single object of respect for a respect-based egalitarian.

This last implication gives rise to the second horn of our dilemma. If we wish to ascribe egalitarian claims to later selves, we must somehow weaken the standard conception of personal identity. Yet advocates of equality of opportunity have two prima facie reasons to resist such a move. First, to the extent that we are less interested in persons understood as temporally unified wholes, we ought to be less interested in freedom than in welfare. We assign people rights to individual freedom because we respect them as temporally extended agents. My freedom concerns the possibilities that I have to perform actions in the future. If my continuity with my future selves is denied, then my freedom to perform such future actions will cease to be a meaningful concern. Instead, what will matter is the set of valuable experiences – that is, the welfare – of any one individual considered at any one time. Second, as Derek Parfit (1984) has argued, if persons are not temporally unified wholes, then not only ought we to be less interested in such persons, rather than in their experiences; we ought also be less interested in the separateness of such persons, which is the very basis of our interest in distributive issues, including issues of interpersonal equality. We ought to promote the relevant good, as utilitarianism prescribes, rather than distributing it equally among biologically continuous individuals.

4 Opacity Respect and Temporally Extended Agency

Despite the concerns raised above in connection with the second horn of the dilemma, I shall now try to show how a more nuanced account of personal identity can help to resolve the ambivalence felt by many egalitarians about starting-gate equality of opportunity – and that it can do so, moreover, in a way that is compatible with a freedom-based account of distributive justice grounded in respect for persons.

Consider the challenge to the standard liberal egalitarian view of personal identity posed by Derek Parfit. According to Parfit, a person’s existence ‘just consists in the existence of a brain and body and the occurrence of a series of interrelated physical and mental events’ (Parfit 1984, p. 211). Our successive selves, understood as time-specific sets of experiences, are connected through such phenomena as memories, intentions, and the continuation of beliefs or desires over time. For example, I am psychologically connected to my former self if I remember having the experience of that former self; and I am psychologically connected to a later self if I intend to do something and that later self does it because he remembers the experience of forming the intention. Connectedness is a matter of degree, in the sense that the number of direct psychological connections between my current self and an earlier or later self can vary. For example, I can remember many of the things I did this morning, but not everything I did: ‘I cannot now remember putting on my shirt this morning’ (Parfit 1984, p. 205). We can say, however, that if there are enough direct connections of this sort occurring between two temporally proximate selves, as indeed is normally the case, then those selves are ‘strongly connected’.

Psychological connectedness is distinct from psychological continuity. The latter obtains in virtue of a series of overlapping chains of strong connectedness. Thus, there can be continuity between two temporally distant selves – say, two selves that are 10 years apart – even if there is no strong connectedness between them, for such continuity depends only on there being a chain of strongly connected proximate selves – say, connectedness from one day to the next, or one week to the next. As time passes, the degree of my connectedness to a specific earlier self can decrease (Parfit 1984, p. 206; Tomlin 2013, pp. 401–3).

Connectedness cannot itself ground personal identity over time, for identity is transitive, whereas connectedess is not. If self A is identical to a later self B, and B is identical to a still later self C, then A is identical to C. By contrast, A might be strongly connected to B, and B to C, without A being strongly connected to C. Continuity, not connectedness is what explains the common belief in personal identity over time, for continuity obtains in virtue of overlapping chains of strong connectedness and is transitive: even if A is not strongly connected to C, A can be continuous with C (Parfit 1984, p. 206). Therefore, we can use the language of personal identity where there is continuity. According to Parfit, however, identity is not important in itself, for the language of identity breaks down in hypothetical situations involving ‘branching’, where a self divides into two but connectedness and continuity still obtain. Parfit believes this conclusion to have normative implications. If C is less connected to A than B is, then C might properly be seen as less praiseworthy or blameworthy for A’s actions than B is, and might therefore be less subject to reward or punishment for those actions.

Most liberal egalitarian theories of distributive justice seem implicitly to treat personhood as what Parfit calls a ‘further fact’ with respect to the empirical facts of organic and psychological continuity, as if a person were an unchanging Cartesian incorporeal substance. He describes his own view, by contrast, as ‘reductionist’, because it implies that there is nothing more to personhood than a set of empirical facts about psychological connectedness and continuity. In what follows I shall depart from the simple ‘further fact’ view of personhood, but attempt to combine some of Parfit’s premises with the account of basic equality set out earlier.

As Christine Korsgaard has noted, Parfit himself makes little reference to temporally extended agency. He includes intentions, along with memories, as a relevant kind of psychological connection, but, as a utilitarian, he focuses for the most part on ‘experiences’, rather than viewing our living of our lives ‘as something that we do’ (Korsgaard 1996, p. 378). Since my focus here is on temporally extended agency understood as a basis for the principle of equal opportunity, I shall assume that the kinds of connection we are interested in are those that exemplify features of agency – that is, those connections resulting from exercises of cognitive and volitional capacities involving the conscious formation, development, recollection, and pursuit of plans and commitments. These connections can be between temporally proximate selves – such as those succeeding one another from day to day or week to week – or between more temporally distant selves – such as selves that are located ten or more years apart.

The next step in my argument consists in noting that continuity, on Parfit’s reductionist account, depends on the presence of a range property. Continuity holds where any two temporally proximate selves in the relevant chain exhibit a sufficient number of direct connections. Exhibiting at least that number of direct connections means falling within a range of connectedness, and that range Parfit calls ‘strong connectedness’. Strong connectedness is therefore a range property. This much holds regardless of whether connectedness and continuity are conceived in terms of passive psychological experiences or in terms of agential notions such as intention, commitment, and the recollection and carrying out of plans.

Now, agential continuity, which involves (let us say) day-to-day or week-to-week strong agential connectedness, is only a part of what is involved in the integration normally achieved by temporally extended agents. In addition to the short-term strong connectedness on which continuity is based, one would expect most temporally extended agents to display a certain degree of connectedness also between more temporally distant selves, as evidenced in the making of more long-term plans, the fact of recalling and honouring more temporally distant commitments, and so on. These further, more long-range connections are of course not normally as numerous as the temporally proximate connections. Nevertheless, we tend to think that, as the number increases, so the person in question qualifies to a greater degree as an agent that is integrated over time. Diachronic integration, then, is another scalar property. And continuity, defined as the existence of chains of strong connectedness at a certain level of proximity, represents, one might say, a minimal threshold of such integration. From these premises it follows that the manifestation of a degree of diachronically integrated agency from the minimum level upwards constitutes possession of yet another range property. Let us call this range property temporally extended moral personality. Temporally extended moral personality is a property possessed by any organism manifesting a series of temporally specific bundles of agential capacities (above a certain threshold), where that series lies within a certain range on a scale measuring levels of diachronic integration. An organism that manifests continuity but only lives from week to week meets the minimal requirement to qualify as a temporally extended moral person: she is a temporally extended moral person, because her level of integration is located within the appropriate range, even though that level is very low.

While echoing the emphasis on agency present in Korsgaard’s critique of Parfit (Korsgaard 1996, 2009), the above account of temporally extended agency dispenses with Korsgaard’s language of ‘unity’ of agency. ‘Unity’ has several possible meanings. Taken literally, in the sense of ‘oneness; being one, single, or individual’,5 unity, when applied diachronically to agency, seems to presuppose either the ‘further fact’ of personal identity mentioned above, or else the ‘full’ or ‘complete’ integration of all the agent’s temporal parts. (I am not sure how to interpret ‘full’ or ‘complete’ integration in this context. Does it entail the strong connectedness of each temporally located ‘self’ with every other?). Taken less literally, as implying ‘due interconnection and coherence of parts’, ‘harmony’ or ‘concord’,6 as it is when applied synchronically to collective agents, unity might denote a range property with a demanding threshold in terms of the required degree of integration. My own account of ‘temporally extended moral personality’ sets the bar much lower and is non-judgemental toward those who, for one reason or another, see little value in achieving high levels of integration (see, for example, Strawson 2004 and Millgram 2014).

Why should we focus on the range property of temporally extended moral personality, thus treating temporally extended agents as equals, and not on the scalar property of diachronic integration, thus treating them as unequals? An answer can again be found in the notion of opacity respect. The account of basic equality expounded in Section 2 was a synchronic account. According to that account, there are certain kinds of relations in which one or both of the parties is required to abstain from taking into account the degree to which the other possesses certain basic agential capacities above a given minimum threshold. Respecting this requirement leads us to focus on the property of having those capacities within a certain range. Extending this reasoning diachronically, in those same kinds of relations we should focus on the range property of temporally extended moral personality because the requirement of opacity respect tells us to abstain from taking into account degrees of diachronic integration above the minimal threshold. In both cases – the synchronic and the diachronic – opacity respect lends moral salience to a range property.

It might be objected that, while opacity respect may be the appropriate attitude to adopt in the presence of a set of basic agential capacities existing at any one time, the same does not apply to the diachronic integration of an agent. The latter involves the exercise of basic agential capacities at various times (recalling commitments, setting ends for the future, making and realizing plans …), rather than the mere possession of such capacities at any one time. It is one thing to possess certain capacities; it is another to exercise them, or to exercise them well. We can judge the ways individuals choose to exercise given capacities without implying that they are deficient in those capacities. The proper object of respect is the possession of certain capacities, not the fact of exercising them in certain ways.7

In the case of temporally extended agency, however, capacity and exercise cannot be distinguished so neatly. Any temporally specific set of agential capacities just is a set of capacities for temporally extended agency, and capacities for temporally extended agency themselves require diachronic integration in the form of agential connectedness. For example: if I only live from day to day or week to week over a given period of time, then the selves occurring after that period will be unable to recall past commitments or to reaffirm, develop, and follow through on inherited plans.8 A similar point can be made if we adopt a backward-looking perspective: if it is true now that my future selves will completely disregard any commitments I make now, or that they will simply be unable, for whatever reason, to recall those commitments, then my current capacities to plan and carry out complex actions are impaired. Therefore, to assess the strength of agential connectedness over time just is, in part, to assess the strength of temporally specific sets of agential capacities. If we accept that opacity respect applies to individuals’ basic agential capacities, then, we must accept that it applies also to degrees of diachronic agential integration.

We know that agential connectedness, both between temporally proximate individuals and between more temporally distant ones, is a matter of degree. Some individuals are more adept than others at forming coherent long-term plans, at recalling plans they have made, at following through resolutely on such plans, and so on. Such individuals are typically more strongly connected at any one time to their more-or-less distant former and future selves. I am suggesting that, in certain contexts, we confer outward dignity on a person by recognizing her as a temporally extended moral person, and that we do this not by assuming the ‘further fact’ view of personal identity or the ‘full’ diachronic integration of her temporally distinct parts or even the high level of such integration implied by the ‘unity’ of agency, but by refraining from taking into account her degree of diachronic integration beyond the minimum threshold of connectedness required to establish agential continuity over her biological life. Integrity, in the sense of internal connectedness and wholeness, is projected onto the agent in order to confer this outward dignity. Korsgaard (2009) says that agents constitute themselves by ‘pulling themselves together’. I am suggesting that, on recognizing a minimal level of such self-constitution, we respect agents by adopting an external perspective and taking their diachronic ‘pulled togetherness’ as given. I am suggesting, moreover, that this attitude helps to explain our view of temporally extended agents as not only separate but also equal.

To reiterate a point made in Section 2, this external perspective need not be seen as appropriate to all human relations. Rather, it has explanatory power to the extent that it is seen as appropriate, for independent reasons, in those relations where we consider the ideal of equality to be applicable. As in the argument of Section 2, there is a strong liberal case for holding it to be appropriate in relations between agents of the state and citizens, and therefore in the context of a political conception of justice.

5 Responsibility for Past Choices and the Right to Distribute Future Opportunities

How might the above account of basic equality and personal identity help to explain people’s mixed feelings about starting-gate equality of opportunity? The account shares with the standard view of personal identity the implication that persons are due respect as temporally extended agents and that, in virtue of their temporally extended agency, they are due equality of opportunity. This recognition of temporally extended agency explains the initial attraction of the starting-gate interpretation of equality of opportunity. At the same time, the account shares with the Parfitian reductionist view the claim that temporally extended agency does not amount to a ‘unity’ of agency based on the ‘further fact’ of personal identity or on ‘full’ diachronic integration (whatever that might mean) or even on a relatively high level of such integration. The pre-theoretical intuitive force of the so-called ‘harshness objection’ to luck egalitarianism may well depend, at least in part, on the plausibility of this second aspect of the above account, and not merely on the welfarist sentiment of compassion implicitly referred to in most diagnoses.

Does the above account also yield a clear and principled normative stance that somehow limits the scope of people’s liability-responsibility under a regime of equal opportunity? One way of moving in this direction would be to say, with Parfit, that a person’s degree of liability-responsibility for past choices should vary in proportion to the degree of her connectedness to the former selves that made the relevant choices. This proposal is endorsed by Patrick Tomlin in the context of a discussion of luck-egalitarianism, precisely on the basis of a Parfitian account of personal identity (Tomlin 2013; see also Navin 2011 and Knight 2015). According to Tomlin, since degrees of connectedness to former selves can, and often do, diminish as the temporal distance from those former selves increases, one’s responsibility for a past choice may well diminish with time; and in proportion as one’s responsibility does so diminish, the current outcomes of a past choice ought to count as brute luck and their costs (or benefits) ought therefore to be socialized.9 But this Parfitian proposal is inconsistent with my account of the relation between equality and personal identity, for it presupposes evaluations of individuals’ degrees of connectedness (above the relevant threshold), whereas my account of basic equality rules out such evaluations, grounded as it is in the requirement of opacity respect for temporally extended persons.

In order to avoid this inconsistency, it might be proposed that, whatever the standard of liability-responsibility applied at time t for choices made at t, that standard ought to be applied at a diminishing rate, uniformly across persons, proportional to the temporal distance from t of the self whose liability is under consideration. In other words, within the context of a responsibility-sensitive egalitarian theory, a person – any person – is to be held less liable at t + 5 than at t for the costs of a choice made at t, and still less liable at t + 10 than at t + 5, with the liability being discounted at the same rate for all. The justification for this interpersonally uniform diminishing rate of responsibility would be the empirical generalization that the connectedness of selves weakens with temporal distance, together with the normative claim that an appropriate standard of responsibility must be one that is fitting for ‘normal agents’, including ‘normally integrated’ agents.

Based as it is on an impersonal fact about human nature, this second proposal would not violate the requirement of opacity respect, and it might permit, for example, the redistribution of resources to the destitute 71-year-old even in the presence of a clear causal relation between her current plight and the gamble she made at the age of 21. However, the proposal would also have morally perverse implications. If you have loaned me €1000 and I have failed to repay the sum by the agreed date, the size of my debt to you will, on the suggestion under consideration, diminish with time. Leaving aside any accumulated interest on the loan, the greater the period of procrastination, the less I will be liable to repay you and the greater the portion of the bill that should be picked up by society as a whole. Moreover, since I am aware of this rule I will have an incentive to hold out as long as possible before repaying the debt. The perversity of these implications suggests the appropriateness of maintaining the same degree of liability-responsibility regardless of the temporal distance between the liable self and the self that performed the relevant actions.

Does such a non-diminishing view of responsibility not depend on the ‘unity’ of agency as discussed above? How can we make sense of the idea that liability-responsibility does not diminish over time even though we know that connectedness with former selves does, typically, diminish over time and even though connectedness is what matters for responsibility? The answer, I suggest, is that we can, and do, think of liability-responsibility as inherited by each successive self from a temporally proximate previous self. It is passed on along the chain of strongly connected, proximate selves, in much the same way as ownership of an entailed property is passed on interpersonally down the generations. For any set of successive, strongly connected selves making up a single biological life, {A, B, C, D … Z}, and a given standard of liability-responsibility applied at the time of A, Z retains full liability-responsibility for an event brought about by A if B inherits A’s liability-responsibility, C inherits B’s liability-responsibility, and so on all the way down to Z. This is why we think that I inherit my earlier commitment to pay you €1000 even if the degree of my connectedness to the self that originally made the promise is weak, and even though connectedness typically diminishes over time. Direct connectedness is not necessary for liability; continuity is sufficient for liability.

How, then, can the starting-gate interpretation of equality of opportunity be resisted? The answer, I suggest, is as follows. Instead of saying that liability-responsibility diminishes over time, we can limit the scope of application of liability-responsibility by limiting the initial opportunities of earlier selves. In other words, compulsory insurance is not ruled out by continuity in the same way as diminishing liability-responsibility is. To see the coherence of this position, we need to compare the following two kinds of normative position implied by starting-gate equality of opportunity:
  • the backward-looking liability-responsibility of an individual for the choices of her past selves. For short, call this ‘responsibility for past choices’;

  • the forward-looking right of an individual to distribute freedom among the set of temporally distinct selves that includes her current self and her several future selves. For short, call this ‘the right to distribute future opportunities’.10

If continuity is sufficient for responsibility for past choices, why should it not be sufficient for the right to distribute future opportunities? To answer this question we need to note two differences in the sources of these two normative positions.

The first difference is that, while liability-responsibility is passed down the chain step by step from A to Z, there is no analogous chain of validity conditions for decisions of A affecting the prospects of Z. A’s right to make such decisions is not exercised through the initiation of some kind of chain of command passing through B, C, D … Y, or through an initial decision that somehow requires the subsequent approval of the intermediate selves B, C, D … Y. Although the actions of these intermediate selves might, as a matter of physical fact, mitigate Z’s plight subsequent to A’s choice (or might indeed not do so), the choice itself is made independently of the wills of B, C, D … Y, despite its impact on the prospects of Z. This first difference suggests that A’s right to distribute opportunities among A, B, C, D … Z might require something more than mere continuity, something more than a mere succession of strongly connected proximate selves A, B, C, D … Z. Although mere continuity between A and Z is sufficient for Z’s liability-responsibility for A’s choices (because such responsibility is passed along the chain), it is not necessarily sufficient to subject Z to the intrapersonally distributive choices of A. Rather, A’s right to deprive Z of various opportunities would seem to require some degree of connectedness directly with Z.

The second difference is that, in the backward-looking case of responsibility for past choices, connectedness is established between two actually occurring selves, whereas in the forward-looking case of the right to distribute future opportunities, the required connectedness would have to be between an actually occurring self and each of an array of alternative possibly occurring future selves. How, then, would direct connectedness between A and Z (more precisely, between A and Z, A and Z*, A and Z** and so on) be established at the time of A’s occurrence? Presumably by reference to a set of clear and specific intentions of A that would have to be formed on the basis of realistic hypotheses both about (a) the probability of there being a Z at all and, in the case of there being a Z, (b) the various options and obstacles that such a Z would face in the many hypothetical scenarios opened up as possibilities by A’s choices, and (c) the various sets of values and sentiments that each of Z, Z*, Z** and so on, might have in those many hypothetical scenarios. This standard of planning looks inhumanly demanding. It is also difficult, from either a third-person or a first-person perspective, ever to be sure that the required degree of connectedness is present, notwithstanding any protestations of A to the contrary.

Assuming the validity of starting-gate equality of opportunity, temporally extended agency is sufficient to grant A an unlimited right to distribute opportunities among A, B, C, D, … Z, subject only to the equal opportunity rights of other temporally extended agents. But for temporally extended agency to have this implication, it seems that it must be based either on the ‘further fact’ view of personal identity or on the dubious empirical claim that agents are ‘fully’ or at least highly diachronically integrated with the array of their future possible selves. The empirical generalization that normal agents are limited in their degrees of integration – in particular, that the connectedness of temporally specific selves diminishes with temporal distance – might therefore provide a basis for limiting the right to distribute future opportunities, even though it provides no basis for limiting responsibility for past choices.

Why, on the view expounded above, ought we to recognize any right at all to distribute future opportunities among one’s future selves? The reason is that we respect persons as temporally extended agents, and we therefore respect them as the originators of long-term plans, including life-plans. On the basis of this respect we should recognize the right of each person to enjoy a substantial measure of freedom at the beginning of her life as an agent. The point of the foregoing argument, rather, is that this consideration lacks the absolute weight implied by the simpler assumption of the temporal ‘unity’ of agency or ‘full’ diachronic integration. Temporally extended moral personality is a range property the normative salience of which derives from the appropriateness of the attitude of opacity respect toward any succession of bundles of agential capacities displaying at least a minimum of diachronic integration. The kind of agent relevant for a theory of equality of opportunity is not the unified agent but the normal agent. Respect for normal agents is compatible with the enforcement of certain interpersonally uniform limits on the right to distribute future opportunities intrapersonally. At the very least, it is compatible with the uniform enforcement of a bare minimum of freedom at various life stages of a person’s life.


  1. 1.

    Relatively pure forms of starting-gate equality of opportunity can be read into Richard Arneson’s luck egalitarian principle of ‘equal opportunity for welfare’ (Arneson 1989) and Hillel Steiner’s left-libertarian principle of equal initial freedom (Steiner 1994).

  2. 2.

    I here assume ‘freedom’ and ‘opportunity’ to be synonymous, and leave open how exactly they are to be characterized. This is not to deny the usefulness, for other purposes, of distinguishing between the two concepts.

  3. 3.

    I part company here with Lowry (2011) and McGeer and Pettit (2015), who argue that liability-responsibility ought to vary in proportion to individual variations in capacity-responsibility. This, at least, within the context of egalitarian normative theory.

  4. 4.

    I expand on this point in Carter 2011b, which is in some senses an embryonic version of this article. In that earlier piece I distinguished between different interpretations of the idea of a person’s freedom being something that extends over the course of her whole life.

  5. 5.

    This is the first definition of ‘unity’ supplied in the Oxford English Dictionary.

  6. 6.

    These are further definitions supplied in the Oxford English Dictionary.

  7. 7.

    I am here assuming that the relevant sense of ‘respect’ is what Stephen Darwall has called ‘recognition respect’, as opposed to ‘appraisal respect’, where the latter would include an assessment of the way people exercises their capacities (Darwall 1977; Carter 2013, pp. 198–99, 201–2).

  8. 8.

    In answer to this example it might be suggested that one can nevertheless distinguish between the capacity to reaffirm, develop, and follow through on an inherited plan and the opportunity to do so, and that it is only the latter that I am denied as a result of my former selves’ failing to exercise their capacities in a coherent way. But this rejoinder is unconvincing, for it is simply not true that it is only my opportunities, rather than my capacities, that are affected by my former failure to make plans. As J.S. Mill reminds us, to exercise one’s capacities is itself to develop them: “The mental and moral, like the muscular powers, are improved only by being used” (Mill 1859/1910, pp. 116–17).

  9. 9.

    More precisely, Tomlin presents two arguments in favour of temporally diminishing degrees of responsibility: the first is based on Parfitian reductionism about personal identity (though without the connection to basic equality contained in my account); the second rests on a particular account of moral responsibility. I cannot here address the second argument, but I believe that the problem I highlight in the diminishing-responsibility view amounts to an objection to both arguments. Admittedly, the problem arises only if I am right that our basic equality is grounded in opacity respect. If this last premise is rejected, the onus will be on the advocate of the diminishing-responsibility view to supply an alternative basis for equality of opportunity.

  10. 10.

    Technically speaking, this right comprises a combination of Hohfeldian powers, on the one hand, and Hohfeldian liberties protected by claims to non-interference, on the other.



For feedback on earlier versions of this article I am grateful to audiences at conferences in Pavia (Ethical Theory and Moral Practice 20th Anniversary Conference, June 2017) and Genova (Italian Society for Analytic Philosophy Mid-term Conference, December 2017), as well as to Patrick Tomlin and two anonymous referees.


  1. Arneson RJ (1989) Equality and Equal Opportunity for Welfare. Philos Stud 56:77–93CrossRefGoogle Scholar
  2. Bou-Habib P (2006) Compulsory insurance without paternalism. Utilitas 18:243–263CrossRefGoogle Scholar
  3. Carter I (2011a) Respect and the basis of equality. Ethics 121:538–571CrossRefGoogle Scholar
  4. Carter I (2011b) Distributing freedom over whole lives. In: Gosseries A, Vanderborght Y (eds) Arguing about justice. Essays for Philippe Van Parijs. Presses universitaires de Louvain, Louvain-la-NeuveGoogle Scholar
  5. Carter I (2013) Are toleration and respect compatible? J Appl Philos 30:195–208CrossRefGoogle Scholar
  6. Darwall S (1977) Two Kinds of Respect. Ethics 88:36–49CrossRefGoogle Scholar
  7. Dworkin R (2002) Sovereign Virtue. The Theory and Practice of Equality. Harvard University Press, CambridgeGoogle Scholar
  8. Fleurbaey M (2005) Freedom with forgiveness. Philosophy, Politics and Economics 4:29–61CrossRefGoogle Scholar
  9. Knight C (2015) Abandoning the Abandonment Objection: Luck Egalitarian Arguments for Public Insurance. Res Publica 21:119–135CrossRefGoogle Scholar
  10. Korsgaard CM (1996) Personal identity and the unity of agency. In: Korsgaard CM Creating the kingdom of ends. Cambridge University Press, CambridgeGoogle Scholar
  11. Korsgaard CM (2009) Self-constitution. Agency, identity, and integrity. Oxford University Press, OxfordCrossRefGoogle Scholar
  12. Lowry R (2011) Blame, reasons and capacities. In: Vincent NA, van de Poel I, van den Hoven J (eds) Moral responsibility. Beyond free will and determinism. Springer, DordrechtGoogle Scholar
  13. McGeer V, Pettit P (2015) The hard problem of responsibility. In: Shoemaker D (ed) Oxford studies in agency and responsibility, vol 3. Oxford University Press, OxfordGoogle Scholar
  14. Mill JS (1859/1910) On liberty. In: Mill JS Utilitarianism, Liberty and Representative Government. Dent, LondonGoogle Scholar
  15. Millgram E (2014) Segmented agency. In: Vargas M, Yaffe G (eds) Rational and Social Agency. The Philosophy of Michael Bratman. Oxford University Press, OxfordGoogle Scholar
  16. Navin M (2011) Luck and Oppression. Ethical Theory Moral Pract 14:533–547CrossRefGoogle Scholar
  17. Olsaretti S (2009) Responsibility and the consequences of choice. Proc Aristot Soc 109:165–188Google Scholar
  18. Parfit D (1984) Reasons and persons. Oxford University Press, OxfordGoogle Scholar
  19. Rawls J (1971) A theory of justice. Harvard University Press, CambridgeGoogle Scholar
  20. Singer P (1993) Practical ethics. Cambridge University Press, CambridgeGoogle Scholar
  21. Steiner H (1994) An Essay on Rights. Blackwell, OxfordGoogle Scholar
  22. Stemplowska (2011) Responsibility and respect: Reconciling two egalitarian visions. In: Knight C, Stemplowska Z (eds) Responsibility and distributive justice. Oxford University Press, OxfordGoogle Scholar
  23. Strawson G (2004) Against narrativity. Ratio 17:428–452CrossRefGoogle Scholar
  24. Tomlin P (2013) Choice, chance and change: luck egalitarianism over time. Ethical Theory Moral Pract 16:393–407CrossRefGoogle Scholar
  25. Vincent NA (2011) A structured taxonomy of responsibility concepts. In: Vincent NA, van de Poel I, van den Hoven J (eds) Moral responsibility. Beyond free will and determinism. Springer, DordrechtCrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media B.V., part of Springer Nature 2018

Authors and Affiliations

  1. 1.Department of Political and Social SciencesUniversity of PaviaPaviaItaly

Personalised recommendations