HUMAN RIGHTS INSTRUMENTALISM

This preliminary section explains how I shall understand instrumentalism as a justificatory strategy. Ronald Dworkin famously argued that when it comes to the justification of basic ethical or political concepts, a theory will give ultimate pride of justificatory place to one of three concepts: ‘it will take some overriding goal, or some set of fundamental rights, or some set of transcendent duties, as fundamental, and show other goals, rights, and duties as subordinate and derivative’.Footnote ⁸ Dworkin's insight is that a practical theory will regard as basic in the order of justification one member of the normative triad goal/duty/right and see the other two as subordinate and derivative.

Dworkin's classification offers a helpful way of characterizing the structure of instrumentalism about human rights. As I shall understand it, an instrumentalist account takes human rights as morally derivative and grounds them as means for the production of a morally basic goal, a good intelligible independently of the concept of rights.Footnote ⁹ As John Tasioulas puts it, ‘we should not accord rights a foundational role in ethical thought: Rights are derived from (certain of) our interests that can be specified independently of the concept of a right.’Footnote ¹⁰ The fundamental justification of human rights is instrumentalist in the sense that human rights are valid in virtue of the fact that they (tend to) produce good states of affairs, the goodness of which is intelligible independently of the concept of rights. For any instrumentalist account, then, human rights have a conditional rather than unconditional validity: human rights are justified on the condition that they meet an external goal.Footnote ¹¹ By contrast, a non-instrumentalist account will be rights-based and will grant rights a ‘foundational role in ethical thought’ by representing them as trumps or as aspects of the inviolable status of persons.Footnote ¹²

Of course, instrumentalist theorists will have important disagreements. Does a single good ground human rights, or does a plurality of goods ground them?Footnote ¹³ And what exactly is the nature of this good or goods?Footnote ¹⁴ What matters for our purpose is that these various accounts share the same instrumentalist structure, for the problems I will highlight pertain to the structure itself rather than to a specific articulation of it.

CATEGORICAL DUTIES

Before assessing whether an instrumentalist account can accommodate the categorical duties thought to correlate with human rights, we need a working notion of a categorical duty.

For the purposes of the argument here we may say that a categorical duty binds necessarily, is pre-emptive or exclusionary, and is non-aggregative.Footnote ¹⁵ To say that a categorical duty binds necessarily is to say that it does not bind contingently. Unlike categorical duties, your reason to purchase eggs is contingent on your having the further end, for instance, of making an omelette. To say that a categorical duty is pre-emptive is to say that it blocks considerations of other kinds as reasons for violating the duty.Footnote ¹⁶ And to say that a categorical duty is non-aggregative is to say that a categorical duty may not be violated in order to prevent a greater number of violations of the same kind of duty.Footnote ¹⁷

Note that this notion of a categorical duty is relatively lax and capacious. For example, it does not commit us to a specific metaethical view about the nature of practical reasons, such as the view that categorical duties, as specifications of the morally obligatory, are a sui generis kind of practical reason.Footnote ¹⁸ And it does not commit us to the stringent normative view that categorical duties must be absolute and exceptionless. Instead, for the purposes of our argument, we can grant that under certain catastrophic circumstances, when we surpass a certain threshold of costs, the duty may be overridden by considerations of other kinds. But within the non-catastrophic threshold, a categorical duty is marked out by its inferential role in practical reasoning, namely, the necessary, exclusionary and non-aggregative character of certain practical reasons.Footnote ¹⁹

The three features of a categorical duty (necessity, pre-emptiveness and non-aggregation) enable us to make three corresponding inferences from the claim that there is a categorical duty to respect a given human right, such as the right to life. First, the claim entails that my duty to respect another person's life binds necessarily and is not contingent on any specific ends I may happen to have, such as perfecting my empathetic disposition. Second, the claim entails that my duty to respect another's life blocks considerations of other kinds from overriding my duty. I cannot fail to respect another's right to life if doing so generated more goodness in the world, such as more welfare or autonomy. Third, my duty to respect another's right to life cannot be overridden by the cumulative weight of the right to life of others. I cannot murder you, for instance, if doing so somehow produced a society more respectful of the right to life of others.

It is important to explain at least a working notion of categorical duties because it is a widely held assumption that human rights correlate with such duties. Non-instrumentalist thinkers like Alan Gewirth and Thomas Nagel, for instance, make it a condition of adequacy of any account of human rights that it be able to accommodate their categorical binding force.Footnote ²⁰ Instrumentalist thinkers also tend to claim that at least some human rights must bind categorically. Rowan Cruft, like Gewirth and Nagel, makes categorical binding force a necessary condition of a right counting as a human right, even if he ultimately defends a version of an instrumentalist account.Footnote ²¹ And John Tasioulas, one of the foremost proponents of an instrumentalist view, upholds the same assumption:

Tasioulas opens his defence of instrumentalism by acknowledging the problem of how ‘we may advance from premises about universal human interests to conclusions about human rights’.Footnote ²³ He argues persuasively that the only way to make this transition work and to bring a genuine human right into view is to explain also how premises about interests can lead to conclusions about categorical duties. Tasioulas's thought, then, is that it belongs to the idea of human rights that they correlate with categorical duties. Conversely, if the stratum of interests failed to generate categorical duties correlated to rights, Tasioulas argues, the stratum of interests would fail to explain human rights. Unless we can explain how human rights correlate with categorical duties of respect, we risk committing a category error, for, as Tasioulas puts it, evaluative judgements of interests belong to the domain of prudence or well-being, while human rights ‘are moral standards that impose duties on others, where the violation of the duty entails wronging someone in particular – the right-holder’.Footnote ²⁴

In sum, non-instrumentalist and instrumentalist thinkers alike tend to assume that human rights correlate with categorical duties of respect.

THE EMPIRICAL CONTINGENCY ARGUMENT

Assuming, then, that human rights correlate with categorical duties of respect, we should ask: can an instrumentalist account accommodate such categorical duties? This section develops the Empirical Contingency argument: instrumentalist accounts make it an empirically contingent matter whether, on any given occasion, others have duties to respect your human rights. And if so, we have reason to answer our question negatively: instrumentalist accounts face fundamental difficulties accommodating the categorical force of duties to respect human rights.

Before developing this argument, let me clarify its point. The point of the Empirical Contingency argument is not that instrumentalist theorists cannot accommodate categorical duties as such. For the purposes of the argument here, we can concede that a classical utilitarian could hold that we have a categorical duty to maximize happiness and minimize suffering. Similarly, the point is not that instrumentalist accounts cannot really justify rights. Instead, the point of the argument is that instrumentalist models face fundamental difficulties accommodating those categorical duties that correlate with human rights. What marks out these categorical duties is that they are not duties to promote a kind of good independent of rights. Rather, these duties are, in the first instance, duties of respect: in the jargon, directed duties owed to others. Yet, for any instrumentalist account, such directed duties, duties owed to others, can only be determined and justified instrumentally. As Nickel and Reidy put it, for an instrumentalist model,

The point of the argument, then, is to show that by taking an instrumentalist approach to the relationship of right (the correlation human right–categorical duty), an instrumentalist model faces fundamental difficulties accommodating the categorical force of such duties.

Let us begin, then, by highlighting a key feature of instrumentalist justification in general: the normative justification of a means is contingent on the extent to which the means is productive of the end. If X is justified only as a means for Y, and if Y is intelligible independently of X, then X is justified contingently, that is, is justified only in so far as X is an efficient means for producing Y.Footnote ²⁶ In turn, this entails that the justification of the instrumental judgement itself will be contingent on empirical circumstances, specifically, on the availability of alternative means of producing the end. For instance, the usefulness of a watch is doubly contingent: it is contingent on the extent to which a watch is a useful means of producing its end of telling time, and it is contingent on the availability of other (and potentially more efficient) means of telling time, such as cell phones and tablets.Footnote ²⁷

Generalizing, we can call this fundamental point about the normativity of the instrumental judgement its Empirical Contingency:

In order to judge that a watch is a useful means of telling time, we must judge not only how well a watch tells time, but also how well alternative means tell time. And that latter judgement presupposes the empirical judgement of what alternative means are available.

Applied to the topic of human rights, Empirical Contingency says that, if human rights are justified instrumentally, the justification of human right R will be contingent not only on its effectiveness in producing a certain goal (e.g. the value of personhood, the protection of urgent interests, the fulfilment of capabilities), but also on the effectiveness of R relative to alternative available means A, B, C, etc. This entails that instrumentalist accounts make it an empirically contingent matter whether, in each case, R is the best means for producing the goal in question. If alternative means were more efficient than R, the logic of the instrumental judgement would warrant dropping R and opting for the alternative instead. If on certain occasions alternative means were more efficient than the given human right, then there would be warrant not to respect and to violate said human right.

If this line of thinking is correct, Empirical Contingency articulates the fundamental difficulty instrumentalist accounts have in accommodating categorical duties. Empirical Contingency suggests that since instrumentalist accounts of any stripe make the justification of human rights conditional on their realization of an external good, human rights cannot have categorical force.

Alternatively put, for instrumentalist models the relationship of right, the duty to respect the human rights of others, matters only indirectly: the duty matters not because it is owed to the other, but rather because it generally promotes some fundamental intrinsic good. But if duties correlative to human rights matter only indirectly, their justification will be contingent and conditional, rather than categorical and necessary.

Let me illustrate this thought with two kinds of human rights that appear to be clear candidates for correlating with categorical duties, the human rights to life and to security of the person correlating with duties not to commit genocide or torture.

Consider, for instance, the Srebrenica massacre of 1995. If the correlation between human rights and categorical duties holds, one would be entitled to claim that General Ratko Mladić violated a categorical duty not to commit genocide against his victims. Can an instrumentalist account accommodate this thought?

The point of concern is that instrumentalist accounts seem committed, in virtue of their justificatory structure, to make General Ratko Mladić's duty not categorical but contingent on whether, on that fateful occasion in 1995, alternative means would be more efficient in producing the grounding end. According to instrumentalist accounts, the justification of the judgement A has a duty not to commit genocide is instrumental: A's duty is contingent on and derivative from more basic goods, such as normative personhood, capacity realization, or the protection of urgent interests. Obviously, A's genocide against the minority population contravenes some of these goods, by diminishing overall normative personhood, capacity realization, or the protection of urgent interests. But suppose that on this occasion the massacre, as odious as any massacre is, was an efficient means of ending the war and thereby avoiding a greater loss of life. If so, an instrumentalist account would have difficulties warranting the view that anyone has a duty not to commit genocide.

In response, one might argue that instrumentalist and non-instrumentalist accounts alike recognize that under catastrophic circumstances a duty not to commit genocide may be overridden. In non-catastrophic circumstances, the duty remains categorical.

Nevertheless, the point of concern is not that under catastrophic circumstances an instrumentalist account would have to override a duty not to commit genocide. Rather, the point of concern is that even when the catastrophic threshold is not surpassed, the instrumentalist would not be able to accommodate the categorical force of the duty not to commit genocide. More generally, if on some non-catastrophic occasions committing a small genocide is most conducive to overall or aggregate goodness, then contravening the duty not to commit genocide would be a more efficient means of producing goodness. If so, then instrumentalist accounts seem committed to the view that even under non-catastrophic circumstances the duty to respect the human right to life and not to commit genocide cannot be categorically binding.

A similar difficulty emerges when reflecting on the right not to be tortured. This case can be even more difficult for instrumentalist accounts, since the circumstances under which torture may be warranted by the good outcomes it would secure can seem to occur more commonly. Once again, the point of concern is not that instrumentalist accounts would break down and warrant torture in catastrophic circumstances. Rather, the point of concern is that instrumentalist accounts would warrant torture even in ordinary circumstances.

So, instead of thinking of the catastrophic case where the torture of one appears required in order to save masses, think of an ordinary case where torture is expected to lead to saving a single child presumably kidnapped by the potential victim of torture. For instance, in 2005 Wolfgang Daschner, deputy chief of police in Frankfurt, threatened Magnus Gäfgen, the presumed kidnapper, that unless Gäfgen revealed where the kidnapped child was, Gäfgen would be subjected to very severe (though not physically damaging) pain.Footnote ²⁸ The kidnapper readily revealed the hiding place, but, as the police discovered, the child had already been murdered. Subsequently, Gäfgen sued Daschner for having threatened him with torture. In its ruling, the court explicitly rejected the idea that the dignity of the suspected kidnapper could be weighed against claims made by the dignity of the kidnapped child.Footnote ²⁹ This is another way of saying that Daschner's duty not to torture Gäfgen is categorical.

Could an instrumentalist theory accommodate the categorical force of Daschner's duty not to torture Gäfgen? It is difficult to see how it could do so. An instrumentalist theory seems committed to weighing the goodness of saving an innocent child's life with the evil of torturing a presumed kidnapper. Since instrumentalist theories would assign a greater weight to the former value, they would seem committed to permitting torture even in circumstances short of catastrophe. The duty not to torture, then, becomes contingent on the empirical fact that, in this instance, not torturing (or threatening to torture) Gäfgen was not the most efficient way of producing goodness. The duty not to torture cannot then be categorical.

At this point, one might object that my argument has relied on the assumption that an instrumentalist theory must be aggregative. But some instrumentalist thinkers reject this assumption. Thus, James Griffin argues that his account of human rights is teleological but not consequentialist.Footnote ³⁰ Similarly, John Tasioulas argues that the choice between surrendering ‘the resistance to trade-offs characteristic of human rights’ and accepting human rights as morally basic is a false one. Instead, he argues that ‘we should not accept his [Nagel’s] premise that the moral logic appropriate to interests is exclusively one of aggregation’.Footnote ³¹ If an instrumentalist theory is non-aggregative, then it would be able to accommodate the categorical duties not to commit genocide or torture.

It is true that I have assumed that an instrumentalist theory has an aggregative structure and that is one important reason why it struggles to accommodate categorical duties. So does an instrumentalist account have the resources to block aggregation?

One reason to find Griffin and Tasioulas's protest unconvincing is that it is not clear how an instrumentalist account can avoid aggregation without taking for granted trade-off resistant rights. And if the account took for granted trade-off resistant rights, it would collapse into a traditional non-instrumentalist account.Footnote ³² Recall, according to Tasioulas, the mark of an instrumentalist theory is that it grounds rights in interests or values ‘that can be specified independently of the concept of a right’.Footnote ³³ Our question is this: if the value can be specified independently of the concept of a right, how can said value block aggregation?

We may glean an answer from a comment by Griffin. Griffin says that the interests of a person must be ‘ring-fenced’Footnote ³⁴ with human rights. The interests of individuals can block aggregation, then, because they are sufficiently important to be ring-fenced with human rights.

But if this is the best answer to our question, it is not compelling. One way to see this is to put the point as a dilemma. On the one hand, if the distance between the grounding value and the ring-fencing rights is not too great, we can say that the interests of individuals are sufficiently important to block aggregation. But, then, what exactly blocks aggregation? What characterizes the unique importance of the interests of individuals? It can seem that the only way the importance of my interests as an individual blocks aggregation with the importance of your interests is by appealing to the notion that my interests are ring-fenced by rights. But if that is the way in which my interests block aggregation, the price of blocking aggregation would be to surrender instrumentalism. The importance of my interests would no longer be ‘specified independently of the concept of a right’, for the very importance of the interests already makes reference to the notion of rights. On the other hand, we could preserve instrumentalism by increasing the distance between the importance of my interests and the protecting rights. But then it is no longer clear how the importance of individual interests by itself is sufficient to block aggregation. If in non-catastrophic circumstances the best way to protect the value of a child's life is to torture a kidnapper, how exactly do the interests of the kidnapper block aggregation? It begins to seem as if accounts appealing to the notion of the importance of interests ‘specified independently of the concept of rights’ simply lack the resources to block aggregation. If so, the objection fails, for it can seem that the only way for an instrumentalist theory to block aggregation is to rely covertly on the notion of rights, thereby collapsing into non-instrumentalism.

Of course, I do not claim to have proven that every instrumentalist theory must be aggregative. My point is only that Griffin and Tasioulas's claim to the contrary is not convincing. In the final section, I will examine in detail three ways in which Griffin argues that a teleological account can show human rights to be resistant to trade-offs. For now, I think we are justified in holding that instrumentalist accounts are aggregative. And if so, they cannot accommodate categorical duties to respect human rights.

Let me strengthen the argument from Empirical Contingency by considering other kinds of cases.

First, one might worry that the argument against instrumentalism trades on the assumption that categorical duties must be absolute. But as I suggested in the previous section, we need not make such an assumption in order to develop the Empirical Contingency argument.Footnote ³⁵

In 2006 the Constitutional Court of Germany considered a law passed in the wake of the terrorist atrocities of September 2001, the Air Safety Law, empowering the state to authorize the shooting down by the air force of a hijacked airliner if there were overwhelming reason to believe that it would be used as a weapon to cause further casualties.Footnote ³⁶ The Court struck down the law, arguing that to shoot down a hijacked plane would be to ‘ignore the status of the persons affected as subjects endowed with dignity and inalienable rights’.Footnote ³⁷ But note that the Court's judgment did not turn on an absolute ban on killing, for it permitted shooting down the plane if it were occupied solely by terrorists. Regardless of how we justify killing terrorists who are about to kill others, e.g. in terms of their forfeiture of the right to life, the point is that there is no obvious inconsistency in holding that the duty to respect the human right to life binds categorically (thereby forbidding shooting down a plane full of innocents) though not absolutely (thereby permitting killing terrorists who are about to murder others).

Our question is whether an instrumentalist account could accommodate the Court's upholding of a categorical duty to respect the human right to life in this kind of case. The Empirical Contingency argument suggests a negative verdict. In this kind of case, if the benefit of saving lives were indeed greater than the cost of losing lives by shooting down the hijacked plane, an instrumentalist theory would have to sanction shooting down the hijacked plane. But if that is so, an instrumentalist account would find it difficult to accommodate the categorical though not absolute duty of the state to respect the human right to life of its members.

Second, moving beyond the cases of the rights to life and against torture and genocide, let us suppose there is a human right to due process. Would the state have a categorical duty to respect this right? Here is one way to formulate the Empirical Contingency argument. Suppose that under typical circumstances the rate of violent crime in a given society could be dramatically reduced if state officials did not recognize a right to due process: the police could summarily arrest and even execute criminals and do so unencumbered by a proper judiciary process.Footnote ³⁸ If this obtained, an instrumentalist analysis would repudiate the warrant for respecting and instituting a right to due process, since recognizing such a right would be more productive of evil than of good. But once the justification of the right to due process is fixed in this way, the duty to respect the right to due process becomes empirically contingent and cannot be categorical.

Third, assume for the sake of argument that there is a human right to political participation. If so, states would have a categorical duty to respect and protect this right. Suppose then that in a given society the level of well-being, capacity realization, or whichever fundamental intrinsic good one prefers, could be better realized if no one were granted a right to political participation. Suppose further that the level of GDP might be more efficiently increased in an authoritarian and centralized rather than democratic government. This would mean that in such a society an instrumentalist account would not recognize a duty to respect the right to political participation, since doing so would be a less efficient means of producing the preferred fundamental intrinsic good. And once the justification of the right to political participation is fixed in this way, the duty to respect the right to political participation becomes empirically contingent rather than categorical.

The reason for considering these other kinds of cases is to see that it is not just the duties not to torture or not to commit genocide that would not be categorical under an instrumentalist account. Rather, the duties to respect other largely uncontroversial human rights, such as the rights to life, due process or political participation, could not be categorical either. Similar cases may be constructed for any other right one takes as a human right.

In sum, my point has not been that an instrumentalist account cannot accommodate any categorical duties or that it cannot justify rights. Instead, the point of Empirical Contingency is that the very structure of an instrumentalist account makes the justification of the directed duties correlative to human rights an indirect, contingent and instrumental matter. As a result, it becomes difficult to see how such duties could be categorical.

INDISPENSABLE MEANS

But this argument can seem vulnerable to the objection that it works only on the assumption of a crude picture of instrumental justification. Indeed, there are two ways of making instrumental justification more sophisticated: strengthening the instrumental necessity of human rights or shifting to an indirect model of instrumental justification. The next two sections develop these more sophisticated models of instrumental justification and show how these models remain vulnerable to the same problem.

The first objection is that I have relied on too crude a picture of instrumentalist justification, for I have conflated the distinction between X being a useful means for the production of Y, and X being an indispensable means for the production of Y.Footnote ³⁹ Human rights are not simply useful means for protecting, for instance, the values of personhood. Instead, human rights are indispensable means for protecting the values of personhood. Human rights are not merely a useful tool; they are an indispensable tool. And once we keep this distinction in view, we should see that at most the empirical contingency argument is effective against the conception of human rights as useful means but is harmless against the conception of human rights as indispensable means. My argument, then, presupposes too crude a picture of instrumentalism and strikes at a straw man.

This line of thought makes a valid point: the normativity of useful means is different from that of indispensable means. We can grant that if X is truly indispensable to Y, the level of contingency in the usefulness of X greatly decreases. We get a stronger sense of the necessity of X for the production of Y. So is this stronger conception of necessity sufficient to accommodate categorical duties?

Nevertheless, upon reflection, we should see that there is a troublesome ambiguity in the notion of an indispensable means. Once we clarify this ambiguity, the same problem remains.

There are at least two senses in which a means may be indispensable to its end: one conceptual, the other empirical. A means is conceptually indispensable to its end when the end is unintelligible independently of the means. For instance, we might think that the human right to security of the person is conceptually indispensable to the end of autonomy because autonomy is unintelligible independently of said human right. Regardless of empirical circumstances, you could not be autonomous without a human right to security of the person. By contrast, a means is empirically indispensable to its end when the end is still intelligible independently of the means. While in some circumstances your autonomy would not require the human right to security of the person, in some empirical circumstances it does.

The difficulty, then, is that once disambiguated, the notion of an indispensable means fails to dodge the problem of empirical contingency.Footnote ⁴⁰ On the one hand, conceptual indispensability is ineffective because it collapses into non-instrumentalism. If the end (e.g. autonomy) is unintelligible independently of specific human rights (e.g. the right to security of the person), then human rights are no longer morally derivative in the relevant sense. Human rights would then violate Tasioulas's condition for an instrumentalist account, namely, that the underlying value be ‘intelligible independently of the concept of rights’. Since there is no conceptual gap between human rights and the end that presumably grounds them, such an account would collapse into non-instrumentalism even if it preserves the language of ‘means’. On the other hand, if the collapse is avoided by conceiving of indispensable means in an empirical way, this simply reproduces the problem I have been articulating. Even if human rights are empirically indispensable to producing their end, this is not sufficient to block aggregation or trade-offs against other kinds of values. For instance, even if the human right to life is empirically indispensable to our preferred fundamental intrinsic good, an instrumentalist account would still find it difficult to uphold the German Constitutional Court's prohibition on shooting down hijacked planes or to prohibit Daschner from torturing Gäfgen to save the kidnapped child.

If this is right, my argument does not strike at a straw man. Indeed, the very notion of an indispensable means supports the argument sketched in the previous section. If the means is conceptually indispensable, then the account preserves the normativity of human rights by abandoning instrumentalism altogether. But if the means is empirically indispensable, then the validity of human rights will be empirically contingent rather than categorical.

INDIRECT INSTRUMENTALISM

The second objection is that I have relied on too crude a picture of instrumentalist justification, for I have assumed that an instrumentalist account must be direct rather than indirect. Mirroring indirect forms of consequentialism, such as rule-consequentialism, we could formulate an indirect form of instrumentalism. The objection would then be that where a simple act-instrumentalism fails to accommodate categorical duties, a more sophisticated, indirect form of instrumentalism succeeds. Let us investigate this proposal. My suggestion will be that indirect forms of instrumentalism suffer from the same defect as direct forms: they make the validity of duties to respect human rights an empirically contingent matter and therefore cannot accommodate naturally the categorical character of such duties.

An indirect instrumentalist holds the view, roughly, that the criterion of rightness is not the good brought about by particular acts, but rather the good brought about by rules actual conformity to which is most productive of good states of affairs.Footnote ⁴¹ The instrumental judgement of goodness now takes for its object not specific acts but rules or rights: rule or right R is justified morally in virtue of the fact that actual conformity to R produces the best available outcome. A direct form of instrumentalism has the familiar disadvantage that it can sanction violating rules when doing so will be productive of better outcomes, thereby contravening intuitive deontic restrictions. An indirect form of instrumentalism promises the advantage that it can accommodate the apparent deontic character of moral rules, such as keep your promises or do not steal. Since what matters to the indirect instrumentalist is the goodness of rules, the fact that on a certain occasion breaking the rule would have better outcomes is not sufficient to ground a duty to break the rule, for having a rule that is (almost) always respected itself can have a better outcome than breaking the rule whenever, on each occasion, it is optimific to do so.

This is familiar territory in normative ethics. Although theorists of human rights have not exploited this avenue as well as they could have, it is worth exploring whether an indirect form of instrumentalism can accommodate the categorical force of duties to respect human rights. Needless to say, the analysis that follows is limited to the viability of an indirect form of instrumentalism as a defence of categorical duties correlated to human rights, rather than as a comprehensive ethical theory.

First, let us sketch the structure of an indirect instrumentalist justification of human rights. Such a view would hold, roughly, that human rights are justified because actual conformity to human rights is most productive of good states of affairs. An indirect instrumentalist view would promise handling the problematic cases I have developed. A world with the rule do not commit genocide or do not torture is a better world than one without such human rights rules, for, arguably, a world without such rules would be one where people suffer more from genocide and torture. For this reason, an indirect instrumentalist would not sanction committing genocide or torturing if, on a certain occasion, doing so would have a better outcome. This may suffice, then, for an instrumentalist to accommodate the categorical binding force of the duty to respect human rights.

Nevertheless, I would like to suggest that there is still reason to doubt whether an indirect form of instrumentalism would succeed in accommodating categorical duties to respect human rights. To see why, consider a classic indirect (or two-tier) justification of rights offered by T. M. Scanlon. Scanlon plausibly argues that, on such a picture, the justification of any moral right would require three kinds of claims. First, it would require ‘an empirical claim about how individuals would behave or how institutions would work in the absence of this particular assignment of rights’; second, it would involve ‘a claim that this result would be unacceptable’; and third it would require ‘a further empirical claim about how the envisaged assignments will produce a different outcome’.Footnote ⁴² On reflection, I shall now argue, the first and third conditions are problematic because they reproduce the problem of Empirical Contingency.

Begin with the third condition. Scanlon's plausible claim is that for an indirect instrumentalist (or, as he calls it, ‘two-tier’) account, moral (or human) rights would be justified by how assigning these rights ‘will produce a different outcome’. Scanlon acknowledges that this is an empirical, rather than a conceptual, judgement. But if the validity of a human right R is contingent on an empirical judgement about the likely outcome produced by acceptance of R, as a conceptual matter, the duty to respect R cannot be categorical. This is because under certain empirical circumstances assigning the rule or right R would have worse consequences than not assigning it. Under certain empirical circumstances R would command our respect, but not in others. And if so, there cannot be a categorical duty to respect R. Let me develop this point in more detail.

Consider the following argument against a human right to democracy by James Griffin.

Griffin's argument against a human right to democracy is simple and powerful: there is no human right to democracy because it is possible in certain realistic though uncommon conditions that a non-democratic regime may realize the values of personhood more efficiently than a democratic one. This view entails the claim that there is no categorical duty to respect the right to political participation. Under certain empirical circumstances R (here, a right to political participation) would command our respect, but not in others.

The difficulty on which I would like to focus does not concern the substantive view of whether there is a human right to democracy. The trouble is that Griffin's form of argument betrays the problem of Empirical Contingency. The form of Griffin's own argument against a right to democracy can be generalized for any indirect instrumentalist account.

Griffin's argument against a human right to democracy gives unintentional expression to a version of the Empirical Contingency argument: there is no categorical duty to respect an alleged human right to democracy because under certain realistic conditions the values of personhood may be more efficiently realized without democracy. The trouble, then, is that this form of argument may be extended to other (if not all) human rights once they are justified instrumentally. Once we justify human rights instrumentally, under certain realistic empirical conditions the value of personhood (or whichever fundamental intrinsic good we prefer) may be better realized by not recognizing the alleged human right in question. And if so, Griffin's argument against a human right to democracy boomerangs against his general instrumentalist justification of human rights: such an account would be unable to accommodate categorical duties. If this line of reasoning is correct, an indirect instrumentalist account ultimately fails to accommodate the categorical duty to respect human rights for the simple reason that it, like its direct cousin, makes the validity of human rights a contingent and empirical matter.

This problem is not simply academic. When we hear of the ‘Asian challenge’ to human rights, for instance, such a challenge manifests a version of the same problem. For example, in the Vienna Conference in 1993, China challenged the Western ‘imposition’ of civil and political rights. China in effect argued that in its stage of historical and economic development, the most effective means of promoting basic human interests is to think of human rights as ‘political aspirations’ rather than as proper rights. Fully recognizing civil and political rights, in China's specific empirical circumstances, would be a hindrance to the welfare (or whichever intrinsic value a theorist prefers) of its people, even if in a later stage of economic development perhaps civil and political rights would have to be recognized. For instance, if it turned out that in China's empirical conditions not recognizing a right to due process would have a better outcome than recognizing one – perhaps because in so doing China would be able to suppress and prevent crime more effectively and thereby maximize the collective's capacity to produce wealth and lift its members out of poverty – then China would not have a duty to recognize its members’ alleged human right to due process. And if so, there could be no categorical duty to respect a right to due process, for such a duty would be empirically contingent.

China's argument exploits the presumed empirical contingency of human rights under both direct and indirect instrumentalist accounts and illustrates how this account makes it difficult to suppose that human rights correlate with categorical duties. If human rights are justified instrumentally and their validity is keyed only to certain empirical circumstances, by indirect instrumentalist lights there is nothing in principle morally wrong with China's argument.

Another way to put this problem is to focus on Scanlon's first condition. Recall, this is the condition that the justification of any moral right requires ‘an empirical claim about how individuals would behave or how institutions would work in the absence of this particular assignment of rights’. But just as the justification of a moral right presupposes an empirical claim about how individuals would behave in the absence of the right, so too this justification presupposes an empirical claim about the extent of compliance with the right or rule in question. This leads to the problem of partial compliance, a familiar objection to indirect versions of consequentialism.

Here is how Peter Railton, himself a consequentialist, puts the point:

Railton's familiar worry is that any indirect version of consequentialism is necessarily unstable because under certain empirical conditions – conditions of partial compliance where the ideal rule is not generally followed – following the rules would have terrible consequences. For instance, even if an ideal code recommends a right to political participation and to due process, recognizing such human rights can have devastatingly bad consequences in conditions of partial or negligible compliance. Similarly, in conditions of extreme non-compliance with human rights, as in the midst of a brutal war, sticking to the rules against genocide or torture may have devastatingly bad consequences, if committing a small and targeted genocide or torturing some prisoners of war could put an end to the war and result in saving many more lives. By sanctioning such bad consequences, any indirect version of consequentialism, by its own lights, should be rejected.

This is a familiar though important problem for rule-consequentialist accounts. Brad Hooker has offered a plausible solution:

In a nutshell, Hooker's proposed solution to the problem of partial compliance is to build into the ideal version of our preferred indirect instrumentalism a proviso that one prevent great harm, such that, under conditions of partial or negligible compliance, we would be let off the hook, as it were, from normally optimific rules.

As a strategy for defending rule-consequentialism, I find Hooker's response ingenious and promising. However, I think Hooker's solution gives expression to the same problem I have been articulating. Hooker's solution effectively claims that, under certain empirical conditions (conditions of partial or negligible compliance), normally optimific rules cease to be binding, for otherwise an indirect instrumentalist would sanction a course of action with ‘devastatingly bad consequences’. But this amounts to a concession that under certain empirical conditions (conditions of partial or negligible compliance) human rights cease to be categorically binding, for otherwise recognizing or respecting human rights would have bad consequences.

Note, furthermore, that conditions of partial or negligible compliance are not always catastrophic conditions enabling us to override categorical duties. For instance, when China fails to recognize a right to due process in order to better establish an orderly and safe society, it operates under conditions of partial or negligible compliance with the right to due process but arguably not under catastrophic conditions. If China's argument that under its conditions not recognizing the right to due process was most conducive to promoting good states of affairs were correct, Hooker's solution would in principle make room for accommodating China's position. But this would be tantamount to the concession that the duty to respect the right to due process of the Chinese is not categorically binding, even in non-catastrophic conditions. In a word, an indirect instrumentalist account would solve the problem of compliance precisely by granting that human rights cannot bind categorically even in non-catastrophic conditions. And that was precisely the point of the Empirical Contingency argument.

Let me summarize these reflections by clarifying their argumentative aim. Our aim has not been to assess the viability of indirect forms of instrumentalism as comprehensive ethical theories. Rather we have considered whether an indirect form of instrumentalism can succeed in accommodating categorical duties to respect human rights where a direct form of instrumentalism failed. Contrary to what may initially appear, I have argued that the shift to an indirect form of instrumentalism does not offer a straightforward way to avoid the problem of empirical contingency. Indeed, I have argued that both forms of instrumentalism ultimately have the same upshot, namely, in making the validity of human rights an empirically contingent matter, they struggle to accommodate the apparently categorical binding force of human rights. The argument here, then, is not meant as a refutation of indirect instrumentalist accounts of human rights. Rather, the argument seeks to show that the standard resources available to indirect consequentialists do not suffice to show in a natural way how an account with such a form could accommodate categorical duties to respect human rights.

A DILEMMA FOR INSTRUMENTALISM

I have argued that instrumentalist accounts of human rights, whether direct or indirect, face great difficulties accommodating the thought that such rights correlate with categorical duties. Thus, we face a dilemma. We must give up either instrumentalism or categorical duties. My main aim in this article was to show that instrumentalists face this dilemma. And if that is so, it casts serious doubts on the frequent assumption that an instrumentalist account can easily accommodate categorical duties. Before I conclude, I want to explore briefly the second horn of the dilemma and show that giving up on categorical duties comes at a high cost.

One might think that the dilemma is not so troublesome, for what really is lost by giving up categorical duties? We could then hold the deflationary view that human rights generate, at most, pro tanto reasons, reasons that speak in favour of respecting, protecting and promoting human rights. These reasons do not bind necessarily, are subject to aggregation and do not override competing kinds of reasons.Footnote ⁴⁶ Such a deflationary view can seem attractive. For instance, it would enable us to sidestep the need to explain how human rights correlate with categorical duties. Instead, we could hold that human rights generate practical pro tanto reasons, perhaps of a specially stringent weight, but not of a categorical kind.

In spite of its initial attraction, I think this deflationary approach comes at a high cost. Rejecting categorical duties or denying their existence is not difficult to do. The real challenge is that in doing away with categorical duties one also risks doing away with human rights. If so, this would be a high cost indeed.

But why would a deflationary view risk eliminating human rights altogether? As we have seen, John Tasioulas, a prominent instrumentalist theorist, insists that in order to establish the existence of a human right we need to explain how human rights generate a ‘special kind of moral reason’. At the very least, this reason must be exclusionary ‘in that it is not simply to be counted in favor of, or against, a certain action, but also neutralizes at least some countervailing reasons’.Footnote ⁴⁷ If the reason we had to respect human rights would not neutralize at least some countervailing reasons, Tasioulas's thought is that we would lose from view the idea of a human right. Similarly, James Griffin, another instrumentalist thinker, formulates the following puzzle: ‘Teleology can fairly readily explain why human rights are not too resistant to trade-offs. But can it explain why, in the first place, they are resistant?’Footnote ⁴⁸ Griffin's thought is that unless a teleological account can explain how human rights are resistant to trade-offs (how they generate exclusionary reasons), then a teleological account will not be able to explain human rights at all. The question before us, then, is whether we need to go all the way to categorical duties in order to preserve a recognizable picture of human rights.

Indeed, in answering his own question, James Griffin offers one of the best arguments I know of in defence of the thought that an instrumentalist account can explain how human rights are resistant to trade-offs without resorting to categorical duties. Griffin's overall strategy is to distinguish the maximizing (or even satisficing) versions of teleology, such as utilitarianism and other forms of consequentialism, which ‘restrict the test of right and wrong to the production of as much good as rationality requires’, from his version of teleology, which is non-maximizing.Footnote ⁴⁹ Griffin then argues that this strategy can explain three ways in which human rights are resistant to trade-offs without resorting to categorical duties.

Griffin's three explanations, then, offer a good test case for examining a deflationary account. In what follows, then, I will argue that none of these three explanations is convincing. The overall form of my argument will be the same: it is not clear that an instrumentalist account can avoid its maximizing versions without surreptitiously relying on the notion of rights and correlative exclusionary duties. Of course, this stops short of a refutation of a deflationary approach. But my aim is not to refute a deflationary approach. Rather, my more modest aim is to show that since Griffin's argument is unpersuasive, we have reason to think that the deflationary approach comes at the high cost of doing away with human rights. So let us turn to Griffin's three explanations.

First, Griffin argues that human rights are resistant to trade-offs because of the ‘great value we attach to our personhood’. Personhood carries such great value that ‘it has a general resistance to trade-offs with welfare’.Footnote ⁵⁰ But if personhood carries such great value, we can use it to explain how human rights are typically resistant to trade-offs without resorting to categorical duties.

However, notice that Griffin himself contrasts two conceptions of personhood, one deontological, the other teleological.Footnote ⁵¹ On the deontological picture, as Immanuel Kant put it, persons have ‘dignity’. To think of the dignity of persons is to think that they have a standing immediately commanding respect. By contrast, the value of things is fungible, exchangeable with that of other things. On the deontological picture, then, morality is not just about promoting the ends that make life good, it is also about respecting the standing of persons, whose value is independent of the promotion of other goods. On the teleological picture, the value of personhood is one value among many enhancing the quality of life. But unlike the deontological conception, Griffin claims that the features of personhood would ‘not, in principle, [be] immune to trade-off with other elements of a good life, such as accomplishment, certain kinds of understanding, deep personal relations, enjoyment, and so on’.Footnote ⁵² While a deontological conception sees personhood as immediately posing barriers to trade-offs with other values, a teleological picture posits, in principle, no such barriers.

Griffin's own contrast suggests a puzzle: if for a teleological picture the value of personhood is subject to trade-offs with other elements of the good life, then how can ‘the great value we attach to personhood’ be a source for resistance to trade-offs?Footnote ⁵³ This presents a dilemma. If one thinks that personhood gives us a source for resistance to trade-offs, then our conception of personhood must be a version of a deontological picture. If so, an instrumentalist model would explain the resistance to trade-offs of human rights by resorting to a non-instrumentalist picture of personhood and rights, thereby collapsing into non-instrumentalism. But if one preserves a teleological picture, as Griffin himself acknowledges, the value of personhood will not be, in principle, resistant to trade-offs. And this would mean that ‘the great value we attach to personhood’ cannot function as the source for the resistance of human rights to trade-offs with other kinds of considerations. That is because the very notion of the value of personhood, understood teleologically, is one consideration to be weighed and traded off against others. So, by Griffin's own lights, personhood will explain exclusionary duties only when personhood is understood in a deontological and non-instrumental way.

Perhaps Griffin's second and third ways of explaining exclusionary duties are more promising. The second way focuses on ‘discontinuities’: ‘some values – an obvious case being our status as persons – are such that no amount of certain other values can ever equal or surpass them’.Footnote ⁵⁴ Human rights generate exclusionary (though not categorical) reasons because the value of personhood is discontinuous with and cannot be challenged by other kinds of values. The third way is ‘appealing to the distinction between “respecting” and “promoting” goods’.Footnote ⁵⁵ While consequentialist versions of teleology focus exclusively on promoting goods, Griffin's non-consequentialist instrumentalism focuses on respecting goods. This is meant to explain exclusionary (though not categorical) duties because the value of personhood is not to be promoted but respected. Do these explanations work?

I remain sceptical, for, like the first explanation, the second and third explanations appear to beg the question. Begin with the second. One can surely claim that the value of personhood is discontinuous with other values, but what exactly makes it discontinuous? What is the form of this discontinuity? Deontological accounts have a ready answer: the value of personhood is discontinuous because it sets a limit on what we may do in the pursuit of other values. But this answer is not available to a teleological account, for otherwise it would collapse into a non-instrumentalist model. So, how can a teleological account regard the value of personhood as discontinuous in the relevant sense without appealing to a deontological notion? Griffin's answer remains terse and uninformative. Surely, on a teleological picture, one can assign different kinds of weights to different kinds of goods, but the discontinuity required by exclusionary reasons is of a distinct kind: the value of personhood overrides other goods under normal circumstances. Yet, this is precisely what a teleological picture of the value of personhood denies. It is not clear, then, how a teleological or instrumentalist picture has the resources to represent personhood as discontinuous in the right way.

Similarly, it is not clear how an instrumentalist picture can block maximizing considerations by treating respecting the good of personhood as basic without surreptitiously appealing to deontological notions. Here, again, we encounter two fundamental notions of respect for persons. For a deontological notion, respect requires, for instance, that we not use people in certain ways; for a teleological notion, respect requires, for instance, that the good of every person have an equal claim upon us, a claim ‘unmediated by any notion of right’.Footnote ⁵⁶ Now, on the one hand, the instrumentalist clearly cannot presuppose a deontological notion of respect as a way of blocking maximizing considerations, for, again, that would amount to a collapse into non-instrumentalism. On the other hand, the instrumentalist must appeal to a teleological notion of respect, but, unlike the deontological notion, such a notion does not have a built-in resistance to trade-offs. If by ‘respect’ one means counting equally the good of persons, then nothing in that notion of respect offers us the resources to block maximization. If so, the distinction between promoting and respecting goods allows Griffin to distinguish consequentialism from a non-maximizing teleology only by presupposing a non-instrumentalist picture of respect for another's rights. The point of concern remains the same: it is not clear that the notion of respecting goods can be robust enough to buttress the resistance of human rights to trade-offs without surreptitiously appealing to a deontic notion of respect. We return, then, to the same puzzle.

Let us take stock. The second horn of the dilemma for instrumentalism involves rejecting categorical duties and taking a deflationary approach to human rights. This approach raises a puzzle: can a deflationary version of instrumentalism explain the resistance of human rights to trade-offs below a certain threshold? The puzzle emerges because if the rights in question would not be resistant in any way, it would look as if we have lost from view the notion of a right and replaced it with a mere consideration for taking one course of action or another. I have considered what I take to be one of the best attempts to solve this puzzle, namely, Griffin's three explanations of how human rights can remain resistant to trade-offs without correlating to categorical duties. But I have argued that none of these explanations is convincing. In a different way, each explanation makes a good resistant to trade-offs only by presupposing a deontic picture of the role of personhood. Since this picture of personhood implicates a notion of rights, each explanation appears able to block trade-offs only by relying on rights. And if so, Griffin's non-maximizing teleology collapses into non-instrumentalism. Of course, Griffin's are not the only possible arguments in defence of non-maximizing instrumentalism. So my argument here is not meant as a proof that every instrumentalist account must be maximizing or that no deflationary instrumentalist approach can illuminate the idea of human rights. Rather, my argument is more modest: one of the best ways of explaining how a deflationary approach can preserve a recognizable picture of human rights appears unconvincing. Unless a better argument is forthcoming, we have reason to conclude that a deflationary approach to categorical duties would also eliminate human rights. If so, there would be a very high cost indeed to endorsing a deflationary approach.