How to Use Someone ‘Merely as a Means’

Abstract

The prohibition on using others ‘merely as means’ is one of the best-known and most influential elements of Immanuel Kant’s moral theory. But it is widely regarded as impossible to specify with precision the conditions under which this prohibition is violated. On the basis of a re-examination of Kant’s texts, the article develops a novel account of the conditions for using someone ‘merely as a means’. It is argued that this account has not only strong textual support but also significant philosophical advantages over alternative conceptions.

1. Introduction

The prohibition on using persons ‘merely as means’ has been one of the most influential elements of Immanuel Kant’s moral theory. For example, it has been invoked in struggles against slavery and other forms of exploitation, and it informs important ethical codes governing medical research on human subjects (Altman 2011: 1–5). It is widely understood to mean that there is an absolute moral limit to what we may do to one another (and to ourselves) in the service of our ends, no matter how desirable or important those ends may be. The prohibition is regarded as clearly applicable in paradigmatic cases such as enslavement or the secret subjection of individuals to dangerous medical experiments. It is widely regarded as impossible, however, to specify the necessary and sufficient conditions for using another person ‘merely as a means’ in a way that does not yield problematic implications for other sets of relevant cases. As a result, the prohibition on using persons merely as means is widely considered unable to provide concrete moral guidance.¹

In this article, I develop a new interpretation of the meaning of the prohibition that yields a clear specification of the conditions under which agents use other persons ‘merely as means’. I first discuss the problems connected with four different readings of the prohibition, found in the literature (section 2). In section 3, I develop the new proposal in three steps. I argue, on the basis of textual evidence, that (1) Kant’s phrase concerns an agent’s ‘using’ others merely as means, rather than ‘treating’ them as such; (2) the phrase articulates a moral constraint on the agent’s practical reasoning, such that the agent ought to make his use of others conditional on their consent, as a matter of moral principle; and (3) the relevant type of consent is genuine actual consent to being used as means to the agent’s ends in a specific way. In section 4, I show that this interpretation avoids the problems associated with other readings, and in section 5 I consider an important objection.

My aim in this article is limited to determining what, according to Kant, constitutes using others ‘merely as a means’. Kant’s conception of using oneself merely as a means merits separate treatment because it requires an examination of the relation between the ‘using self’ and ‘used self’, including a consideration of the sense in which one can give or deny consent to oneself.² Furthermore, I do not discuss Kant’s justification of the claim that it is indeed morally impermissible to use others merely as means, because doing so requires an article of its own. I must therefore also defer discussion of objections to Kant’s claim that using others merely as means is absolutely prohibited. Here my focus is on the prior issue of the meaning of the phrase and the associated prohibition.

2. Problems with Current Accounts of ‘Using Someone Merely as Means’

We often use others as means to our ends. To mention one of Kant’s examples, you may use a mason as a means to build a house, and a mason may use you as a means to obtain money (L-NR, 27: 1319). Not all forms of using others as means are morally permissible, however. Kant argues that using a person ‘merely as means’ is morally prohibited.³

The classic formulation of this injunction is the Formula of Humanity in the Groundwork for the Metaphysics of Morals:

So act that you use the humanity, whether in your own person or in the person of any other, always at the same time as an end, never merely as a means. (G, 4: 429)⁴

Strictly speaking, the prohibition mentioned at the end of the formula concerns the use of the ‘humanity in a person’. By ‘humanity’ Kant here understands that which distinguishes humans from other animals, and this is their capacity to act in accordance with rational principles and to set ends independently of inclination (G, 4: 428–31, 434; CPrR, 5: 87).⁵ This capacity is inseparable from the other elements that together make up a human being, which makes it possible for Kant to speak of using ‘a human being’ or ‘a person’ instead of ‘the humanity in a person’ – just as I can say, when using one particular program on my computer, that I am using my computer. Kant also asserts, on similar grounds, that someone who uses a part of another person’s body can be said to use the person (Eth-C, 27: 386–7; cf. MM, 6: 279). Hence he often speaks of ‘using a person merely as a means’ or ‘using another human being merely as a means’ (G, 4: 429), and in this article I shall do the same.

Kant illustrates the prohibition on using others ‘merely as means’ with the example of an agent who contemplates acting on the following maxim:⁶ ‘when I believe myself to be in need of money, I shall borrow money and promise to pay it back, even though I know that this will never happen’. He explains:

[H]e who has it in mind to make a false promise to others will see at once that he wants to make use of another human being merely as a means, without the other at the same time containing in himself the end. For he whom I want to use for my purposes by such a [false] promise cannot possibly consent to my way of behaving toward him and so himself contain the end of this action. (G, 4: 429–30)

This comment is brief, but it clearly suggests that the prohibition against using others ‘merely as means’ should be understood in terms of a consent requirement. It is less clear, however, how this requirement should be specified more precisely. Is the problem that the lender has not given consent to the agent, that he could not rationally give consent, or that he does not have the option of giving or withholding genuine consent in the first place, since he is being deceived? Does the relevant consent concern the agent’s action, his maxim or the use he wants to make of the other? In other words: what is the relevant kind of consent, and what is the proper object of consent? Let us first consider four different answers that can be found in the literature.

2.1. Actual Consent to Being Used, in a Particular Manner, as a Means to the Agent’s End

Kant’s statement that the false promisor uses the lender without the latter ‘at the same time containing in himself the end’ could be taken to indicate that you use others merely as means if and only if you use them as means without their actual consent to being used, in this way, as a means to your end. If a lender agrees to lend his money, on the basis of a borrower’s promise that he will repay the loan, then he ‘simultaneously contains the borrower’s end’ in the sense that he has adopted the borrower’s end, that is the borrower’s end of temporarily getting to use his money. If the promise is false, however, then he does not contain the quasi-borrower’s end, since he is not aware of the latter’s real end.

In addition to making good sense of Kant’s example, the actual consent reading also seems to be textually supported by passages outside the Groundwork. According to the Metaphysics of Morals, a loan, as such, requires the simultaneous declaration of will by both parties, and this includes the actual consent of the lender to the borrower’s using his money (MM, 6: 271–86, esp. 285). Another case is the use of servants, which, as Kant states repeatedly, requires the servants’ actual consent (e.g. L-NR, 27: 1319, MM, 6: 359-60; I discuss these examples in more detail below).

When the criterion is read as only requiring actual consent, however, it is widely regarded as unsatisfactory. A first cause for concern is the fact that someone’s actual consent may be spurious, for example when it is the result of manipulation or deception (O’Neill 1989: 106–12). If an agent produces consent in these morally impermissible ways, then surely the other’s (spurious) consent does not show that the agent is not using the other merely as a means. This worry regarding spurious consent is no ground for rejecting an actual consent criterion, however, for it can be put to rest – at least in principle, even if perhaps not fully in practice – by spelling out the requirements for genuine actual consent and restricting the relevant consent to the latter.

Other authors claim that on an actual consent reading Kant’s criterion rules out too much. An example discussed in the literature is that of a collapsed jogger in a park who is neither conscious nor breathing. Since she is unconscious, the jogger cannot consent to CPR treatment (and it is tacitly assumed that she is not carrying a living will). Some commentators have argued that, on the actual consent interpretation of Kant’s criterion, therefore, you would be treating her merely as a means if you resuscitated her to save her life. Understandably, this result seems implausible (cf. Kerstein 2009: 174; Formosa 2017: 95, 99–100).

The actual consent reading also seems to rule out too little, however. Imagine a scenario in which a genocidal dictator selects several individuals from the ethnic group she is annihilating, in order to subject them to dangerous medical experiments that promise to yield valuable medical insights that would benefit the rest of humanity. Now suppose that one of the selected individuals happens genuinely to consent to the treatment — say, a radical act-utilitarian who is convinced of the experiment’s overwhelming benefits for large numbers of humans in the long run and who believes that these benefits vastly outweigh his own agony. The genocidal dictator does not care one bit whether any of the selected individuals give consent, and she assumes they do not. When she hears of the act-utilitarian’s consent, she laughs with contempt. It sounds odd to say that she uses all of these individuals except the act-utilitarian ‘merely as a means’. After all, she treats each of them in the same way, for the same reason, in the service of the same end and regardless of whether they give consent. As a result, it is hard to see how the act-utilitarian’s consent could make a major moral difference in the description of the dictator’s manner of acting. It seems that his consent is not appropriately related to her manner of acting for it to affect the moral status of her way of using him.

Similar worries can be raised concerning the use of a person who consents to being enslaved or the use of an enslaved person who consents in a particular case.

2.2. Possible Rational Consent to Being Used, in a Particular Manner, as a Means to the Agent’s End

A second option is to take the relevant type of consent to be possible rational consent to being used, in a particular manner, as a means to the agent’s end. On this reading, you use others merely as means if and only if you use them in a way to which they could not rationally consent. You do not do so if you use them as means in the service of what can be rationally chosen or what is rationally required, even if this involves overriding their de facto dissent.

Derek Parfit’s reading of Kant seems close to this view, since he reads Kant as defending the principle that it is ‘wrong to treat anyone in any way to which this person could not rationally consent’ (Parfit 2011: vol. 1, 220). Parfit claims that on Kant’s principle you would not be using a heavy man merely as a means if you unilaterally decided to push him off a footbridge in order to stop a runaway trolley from killing five others. After all, Parfit argues, the heavy man could rationally consent to being killed as a means to saving the five, given the overall consequences: a net saving of four lives. Parfit considers this alleged result to be a reason for rejecting Kant’s principle (212–29, esp. 220–1).

The most obvious problem for this reading of the Formula of Humanity is its lack of textual support. Kant does not argue anywhere that it is sometimes morally permissible to kill, to steal or to make false promises as long as the victim could rationally consent. Parfit relies solely on Kant’s comment, in his explanation of the false promising example, that the lender ‘cannot possibly’ contain the promisor’s end (Parfit 2011: vol. 1, 177, 179, 181).⁷ This comment does not specify the type of impossibility at issue, however, and it is perfectly compatible with the actual consent interpretation above, as well as with the other readings discussed below. Thus, Kant’s comment that the lender ‘cannot possibly’ consent offers no textual support for the possible rational consent reading as such. Moreover, Kant’s categorical rejection of certain types of action, such as his unconditional condemnation of theft, clearly runs counter to the possible rational consent interpretation. It is not surprising, therefore, that most Kantians and Kant scholars reject rational consent interpretations of what it is to use someone ‘merely as a means’ (e.g. Kerstein 2013: 68–72; Nyholm 2015: 125–35; O’Neill 1989: 109; Pallikkathayil 2010: 120–4).

2.3. Possible Consent to the Action

A third option is to explicate the meaning of Kant’s phrase in terms of the in-principle impossibility of (actual) consent. Here the idea is that you use others merely as means if and only if it is impossible for them to consent to your action, given the nature of your action.

This is the sense in which Christine Korsgaard analyses Kant’s comment that the lender in the Groundwork example ‘cannot possibly’ contain the false promisor’s end. Korsgaard explains that the lender clearly cannot possibly consent to what the false promisor is really doing, since he is being deceived. And suppose the lender sees through the false promise: then it is no longer possible for him to accept it as a promise – at most he might pretend to (Korsgaard 1996: 137–40). Korsgaard argues that ‘[t]he action is one that by its very nature is impossible for the other to assent to’ and concludes that the agent uses the lender merely as a means (139; emphasis added). The ‘criterion for judging whether you are treating [another] as a mere means’ is ‘whether another can assent to your way of acting’ (1996: 139; emphasis added). She emphasizes that the focus is not on the absence of consent but on its impossibility: the problem is not ‘that the other person does not or would not assent to the transaction or that she does not happen to have the same end I do, but strictly that she cannot do so: that something makes it impossible’ (138).

When construed in this third way, however, the scope of Kant’s phrase seems too narrow. There is an important class of cases of violence that seem to involve the use of others merely as a means, although the nature of the action does not make it impossible, for the person who is used, to consent. Consider, for example, the action of using a healthy person’s organs, thereby killing him, in order to save the lives of five others. Does this constitute a case of using the person merely as a means? On Korsgaard’s account, the question here is not whether the person does or would consent, but ‘strictly’ whether the person can consent given the nature of the action. In contrast to the false promising example, here the nature of the action does not make it impossible for the victim to consent. If, in a spirit of radical altruism, he did consent, the agent’s action would still be properly described as using his organs to save five others, thereby killing him. It follows from Korsgaard’s account of the prohibition, therefore, that organ harvesting does not qualify as using the victim ‘merely as a means’, not even when the victim does not or would not consent.⁸ Yet passages elsewhere, such as Kant’s discussion of using another person’s sexual organs (MM, 6: 278) and his moral condemnation of removing one’s tooth for transplantation into the mouth of another (6: 423), clearly suggest that on Kant’s understanding of the notion, organ harvesting should count as using someone merely as a means – most certainly⁹ when a person does not or would not consent.¹⁰

2.4. Possible Consent to the Agent’s Maxim

A fourth option is to read Kant’s criterion as requiring possible consent to the agent’s maxim. On this account, you use others merely as means if and only if they cannot consent to your maxim.

This reading can be found in the work of Onora O’Neill and Thomas Hill. O’Neill claims that in the false promising case, ‘the victim of deceit cannot agree to the initiator’s maxim, so is used’ (O’Neill 1989: 113).¹¹ She writes: ‘To use another as mere means, as Kant sees it, is to act on a maxim that the other cannot also adopt. This amounts to acting on a maxim that one cannot at the same time will as universal law’ (138; original emphasis). The focus, on her account, extends to anyone who is ‘involved in some way’ or ‘affected’ by your action (139, n.), and the relevant question is whether they can agree to your maxim. Hill defends a similar view, claiming that it is ‘not relevant’ whether they happen to adopt your end (Hill 1992: 45).

This fourth reading of Kant’s phrase faces difficulties too, since its scope seems too wide. Many actions that do not involve any use of others at all would qualify as ‘using’ others ‘merely as means’. Suppose, for example, that you have a maxim of never helping anyone in need (cf. G, 4: 423) and that you let me drown while I am crying out for help, although you could easily save my life. In terms of this fourth reading of ‘using merely as a means’, you would be using me merely as a means if you let me drown. After all, I am clearly affected, and since I want help I cannot agree to your maxim of never helping. You are not using me as a means, however, when you refuse to help, and it makes little sense to say that you are using me ‘merely as a means’ when you are not even using me ‘as a means’. Kant indeed explicitly denies that the egoist in the fourth Groundwork example, who refuses to help anyone in need, uses this person merely as a means (4: 430). So, on the fourth reading of the criterion, its scope seems too wide.

2.5. Denying the Problem or Revising the Principle

As a result of the difficulties associated with these four interpretations, many authors view it as impossible to provide a satisfactory reading of Kant’s prohibition, but some deny that this is a problem. Some commentators downplay the importance of those few words at the end of the Formula of Humanity: ‘never merely as a means’. Allen Wood, for example, asserts that ‘far too much’ is made of the claim that we must never treat humanity merely as a means (Wood 2008: 87). He maintains that it is impossible to derive any concrete duties on its basis and that Kant’s claim actually ‘plays no role whatever’ in the content of the formula (Wood 1999: 143).

The assertion that Kant’s claim plays no role is hard to square, however, with his explicit division of his four Groundwork examples into two pairs. As Kant explains it, the first pair includes cases in which the agent contravenes the humanity in a person by using it merely as means; the second pair includes cases in which the agent only fails to promote humanity as an end in itself (4: 429–30). If we follow Wood and read the formula simply as the injunction that we ought always to promote humanity as an end in itself, then this division becomes unintelligible. In each of the four examples the agent fails to promote humanity as an end in itself, so in that regard there is no relevant difference between the two pairs.

Moreover, Kant repeats the prohibition on using persons ‘merely as means’ many times, also in later works, and he appeals to it for the purpose of identifying concrete moral duties and prohibitions. In the Groundwork, he offers not only the example of promising falsely to repay a loan but also that of ‘assaults on the freedom and property of others’ (4: 430). In the Doctrine of Right, he discusses cases in which contracts are required to avoid using others merely as means (MM, 6: 359-61), and in the Doctrine of Virtue he explicates the prohibition on using others merely as means in terms of a specific duty of virtue, namely the duty of respect for the humanity in the person of another (6: 450, 462). In sum, Kant appears to attribute more significance to the prohibition than Wood claims he does.

This still leaves open the possibility, of course, that the prohibition does indeed play an important role in Kant’s work but that the associated problems cannot be solved. This is the view defended by Samuel Kerstein, the author of the most extensive discussion of the notion of ‘using merely as a means’ to date (Kerstein 2013). On his view, Kant’s prohibition, in its original form, has implausible implications. He recommends revising and ‘de-absolutizing’ it, such that it serves as a pro tanto principle that can be ‘outweighed by some other moral reason’ (Kerstein 2009: 179, cf. 2013: 55).¹² Even without further discussion of the merits of Kerstein’s own proposed revision, it is clear that it constitutes a significant departure from the absolute prohibition articulated in Kant’s original formulation (never merely as a means).

In the remainder of this article, I argue that a new look at the texts suggests a novel account of Kant’s phrase, an account that avoids the problems associated with the interpretations mentioned above and gives the prohibition on using others ‘merely as a means’ the status of an absolute proscription with determinate content. The interpretations mentioned above explain the prohibition in terms of whether the persons who are used (or treated or affected) do, can or could consent. The interpretation developed below, by contrast, focuses the attention on the agent’s practical reasoning and its proper relation to the consent of the person who is used. I argue that Kant considers the actual consent of someone who is used as a means to be a necessary but not a sufficient condition for the agent to avoid using this person merely as a means.

3. A New Proposal: An Agent-Focused Actual-Consent Account

I defend the following account of what constitutes using another merely as a means:

An agent uses another person merely as a means if and only if (1) the agent uses another person as a means in the service of realizing her ends (2) without, as a matter of moral principle, making this use conditional on the other’s consent; where (3) by ‘consent’ is meant the other’s genuine actual consent to being used, in a particular manner, as a means to the agent’s end.

I shall clarify the three elements of this account in turn.

3.1. Using as a Means

The first condition without which we cannot properly say that an agent uses a person ‘merely as a means’ is that the agent indeed uses the person as a means to her end. It is trivial but nevertheless sometimes overlooked: if an agent does not use a person as a means to her end at all, the agent cannot be using the person merely as a means.

Importantly, Kant’s own terminology in the Formula of Humanity and related discussions is that of ‘using’ someone (merely) as a means ([ge]brauchen, G, 4: 429–30). In current discussions of the phrase – including those mentioned in section 2 above – however, the relevant relation is often described as that of ‘treating’ persons (merely) as means. This notion has a broader scope, because it also includes regarding persons as being of potential use, or regarding persons in the way in which one regards things, without actually using them.

I suspect that the introduction of the terminology of ‘treating’ is due to several influential but infelicitous English translations. H. J. Paton and Lewis White Beck, two of the most important Kant scholars of the mid-twentieth century, translated Kant’s Formula of Humanity as a prohibition on ‘treating’ humanity merely as means (Paton: ‘Act in such a way that you always treat humanity …’; Beck: ‘Act so that you treat humanity …’).¹³ They likely did so because ‘using … as an end’ sounds awkward. Yet this is what Kant writes, and he repeats the phrase in the Critique of Practical Reason and the Metaphysics of Morals even more bluntly. There he claims that any agent is morally constrained never to use a subject merely as a means but always ‘simultaneously to use this as itself an end’ (CPrR, 5: 87) and that a human being must ‘always simultaneously be used as an end’ (MM, 6: 462). The locution ‘using as an end’ is probably best understood as meaning ‘using qua end’ or ‘in accordance with its standing as an end’, that is, as indicating that one ought to use a human being in a way that is consistent with the latter’s moral standing as an end in itself. In a similar vein Kant also speaks of ‘the use of [a being] as a person’ (6: 276), rather than as a thing.

The infelicitous translation by Paton and Beck has substantial implications. Some of the difficulties currently associated with Kant’s Formula of Humanity stem directly from taking the relevant phrase to be ‘treating merely as a means’ rather than the narrower notion of ‘using merely as a means’.¹⁴ Take, for example, the case in which a beneficent bystander gives CPR to an unconscious jogger in a park. As mentioned above, some authors claim that the bystander would be ‘treating’ the jogger ‘merely as a means’ since, being unconscious, she cannot give consent. But when Kant’s prohibition is read literally, as concerning cases of using others, then the case of the unconscious jogger causes no problems. Suppose you are a good Kantian and you are acting on your maxim of helping others in need, a maxim you have adopted from duty. In this scenario your end is to save the jogger’s life, and the means by which you realize this end is your giving her CPR. You are not using the jogger as a means, and hence by implication you are not using her merely as a means.¹⁵ (The example is not that of an agent who is roaming the parks in search of people to whom he could credibly give CPR just so as to become famous as a life-saving hero.) In short, the worry that resuscitating the jogger violates Kant’s prohibition is an artefact of an unfortunate translation.

Given the crucial importance of the phrase, let us consider the notion of using someone as a means, before focusing on the subset of cases in which such use constitutes using a person merely as a means. What constitutes using another person as a means? Importantly, we cannot establish whether you are doing this by determining whether the other person happens to contribute causally to the realization of your ends. Others may contribute to the realization of your ends even when you do not use them as means, and even when you do not act at all. If I, of my own initiative, buy food for you while you are sleeping because I know you want food, then I satisfy one of your ends, but you are not using me as a means to get food. You are asleep; I am the agent. To count as using me as a means, you need to be acting in a certain way.

Your action can be described from an external narrator’s perspective or in terms of your own practical reasoning, so we need to determine which of these perspectives is the one adopted by Kant in this context. To illustrate the difference, let us suppose someone is at a music festival, where it has gotten dark. She wants to rest, so she sits down on what she believes to be a large bag, to avoid sitting in the mud. In reality, she is sitting not on a bag but on a man who has drunk himself into a coma. An external narrator might say that she is using the comatose man as a means to sitting comfortably. But in terms of the woman’s own beliefs and practical reasoning, this is not what she is doing. She believes that she is using a bag. Unbeknownst to her, this happens to be a man.

For the purpose of determining whether you qualify as using another person as a means in Kant’s sense, we need to describe the action by reference to your own perspective. In the Formula of Humanity, Kant addresses you, the agent, directly (‘So act that you use …’): you are told to critically assess your own action principles (G, 4: 429–30). Moreover, when discussing his four examples in the Groundwork, explaining how the agent in the first two cases would use a person merely as a means, Kant takes as his point of reference what the agent ‘wills’ and ‘has in mind’ and which ‘principle’ he would be acting on (G, 4: 429–30). He criticizes the agent for failing to take into account, in his practical reasoning, the fact that rational beings always simultaneously ought to be valued as ends in themselves. All of this indicates that the relevant point of reference is not the agent’s observable behaviour as described from an external perspective, but the action as conceived in terms of the agent’s own practical reasoning. I shall therefore assume the following: an agent uses another person as a means if and only if, in the agent’s practical reasoning underlying the action, the other serves to realize or promote the agent’s ends.

3.2. Using ‘Merely’ as a Means: The Consent Condition

Not every case of using another as a means constitutes a case of using the other merely as a means. We need a second condition to identify the proper subset of cases. This second condition is the following: an agent uses someone merely as a means if and only if the agent uses someone as a means without having made such use conditional, as a matter of moral principle, on the used person’s consent. In the next section (3.3), I provide the necessary textual support for the claim that the relevant consent is best understood as actual consent. In this section, I focus on the role of consent as a limiting condition.

Recall that whether an agent is ‘using’ another person is determined by reference to the agent’s practical reasoning. The action principles from which agents reason practically are, in Kant’s terminology, the agents’ ‘maxims’. Maxims are the general premises agents use in their practical reasoning, that is, the principles on the basis of which they act (G, 4: 400, n., and 421, n.; CPrR, 5: 19). The notion of a maxim is a familiar element of Kant’s ethics, but because he does not mention maxims explicitly in the Formula of Humanity their importance is often overlooked in accounts of ‘using merely as a means’.

Kant makes it very clear that the prohibition on using persons merely as means articulates a maxim-related constraint. He rephrases the requirement articulated in the formula as follows:

[T]he formula [of humanity] says: that the rational being, as an end by its nature and hence as an end in itself, must serve in every maxim as the limiting condition of all merely relative and elective ends. (G, 4: 436; emphasis added)

He repeats this idea by stating that every rational being must simultaneously ‘count as an end in itself in your maxim’ (G, 4: 437–8) and by saying that rational beings ‘must be made to underlie all maxims of actions’ as the ‘highest limiting condition in the use of all means’ (4: 438). There are subtle differences between these formulations, and Kant does not make his view entirely clear. The first two passages might seem to suggest that the condition should be stated explicitly in each maxim. Yet this does not actually appear to be what is required by the condition that rational beings ‘serve’ and ‘count’ as ends in themselves in your maxims. Kant does not include the condition explicitly in his own examples of morally required maxims, nor does his reformulation in terms of the condition having to ‘underlie’ one’s maxims suggest that it should be stated expressly in the maxim. His idea, rather, seems to be that the condition is to function as a general rider on one’s practical reasoning, imposing a general moral constraint on one’s use of others as means to one’s ends.

Kant explains the nature of this constraint in terms of the consent of the person who is used as a means. He explicates the meaning of the unusual phrase ‘using as an end’ by contrasting it with ‘using as a thing’. Whether I use a thing that I own is entirely up to me. By contrast, human beings are persons, they exist not merely as means to my ends, but they are ‘ends in themselves’ (G, 4: 428–9). They morally cannot be owned and used as property; this would be slavery. Persons have freedom, and the freedom of others ‘limits’ mine (L-NR, 27: 1335). Whether I may use a person is therefore not entirely up to me. It depends on whether the other gives consent.

This suggests that the constraint should be understood in terms of the following moral principle: given the moral status of persons as ends in themselves, it is morally impermissible for one person to use another as a means, unless the other gives consent. If the other does not consent, then the agent must find morally acceptable means, postpone his pursuit of the end or give up the end (unless it is a morally obligatory end).

The distinction between using others merely as means and using others as means but not merely as means can now be stated as follows: you use others merely as means if and only if you use others as means to your ends without making your use of them conditional on their consent. By contrast, you use them as means but not merely as means if and only if, as a matter of moral principle, you make your use of others as means to your ends conditional upon their consent.

3.3. Actual Consent

The third element of the proposed account of ‘using merely as a means’ is the claim that the relevant consent is actual consent. Kant’s explanation of the false promising example in the Groundwork is notoriously brief, but a closer look at this example and other relevant texts confirms that the required type of consent is actual consent.

Recall that Kant starts his example in the Groundwork by stating that the person who contemplates making a false promise to repay a loan ‘will see at once that he wants to make use of another human being merely as a means, without the other at the same time containing in himself the end’ (4: 429, emphasis added). By introducing the case as one in which the other does not simultaneously contain the agent’s end, Kant presents it as a case in which actual consent is lacking. In the second sentence he explains why the other does not simultaneously contain the end, namely because he cannot possibly contain it: ‘For he whom I want to use for my purposes by such a [false] promise cannot possibly consent to my way of behaving toward him and so himself contain the end of this action’ (4: 429–30). After all, he is being deceived as to the agent’s real end.

In the literature, this second sentence is often taken to be Kant’s primary statement of the relevant criterion. This explains the prevalence of the different ‘possible consent’ accounts discussed above. But Kant in fact starts the example by pointing to the absence of actual consent. He subsequently mentions the impossibility of consent as an explanation of its absence. Thus, the relevant type of consent in his example is actual consent.

Other texts similarly reveal that Kant understands the Formula of Humanity as requiring actual consent. In the Feyerabend Lectures on Natural Law, the transcript of a course he taught during the months in which he was writing the Groundwork, he is reported to have said: ‘When I make a contract with a servant, he must also be an end … and not merely a means. He must also will it. (Er muß auch wollen; L-NR, 27: 1319; emphasis added)

In the Metaphysics of Morals, we find many other passages in which Kant stresses the importance of actual consent. Here he again emphasizes that an agent’s use of servants requires their actual consent (6: 283). It requires a contract, which involves a simultaneous declaration of will (6: 272).¹⁶ Servants retain the right to revoke their consent and terminate the contract (6: 283), but as long as the contract remains in force, the employer can ‘make direct use of a person … as a means to [the employer’s] end, … without infringing upon his personality’ (6: 359–60).¹⁷ The requirement of actual consent applies to all cases in which one person wants to make use of the powers or property of another (6: 285, cf. 271–2). Importantly, Kant’s loan example in the Groundwork is a case in point. A loan is not a real loan (but a case of misappropriation) if the lender does not give genuine actual consent to lending the money. The lender ‘must also will it’.

In addition, Kant’s servant and loan examples clarify that the object of the required actual consent is not the agent’s maxim. An employer’s maxim may well be a maxim that a potential employee wholeheartedly can or does endorse – for example the maxim ‘when hiring employees, I shall offer them fair terms of employment’ – but unless the potential employee actually agrees to take the job, the employer cannot permissibly make use of his services. The employee ‘must also will it’. The prospective employee may have reasons of his own to prefer another job elsewhere: perhaps the pay is better or the work more agreeable. In other words, for us to avoid using others merely as means, it is not sufficient that those whom we use as means can or do consent to our maxims. We need to secure their genuine actual consent to our using them as means to our ends in a specific way. The relevant object of consent, then, is not our maxim but the use we want to make of the other as a means towards our end.

Its focus on the agent’s practical reasoning makes the proposed account fundamentally different from the actual consent reading mentioned first in section 2, to which I shall now refer as the ‘bare actual consent’ account. There, the criterion governing whether an agent uses someone merely as a means was phrased solely in terms of a condition on the side of the person used: their actual consent. By contrast, on the agent-focused reading proposed here, the criterion is phrased in terms of a condition on the side of the agent. On this reading of Kant, it is not morally sufficient that others happen to consent; the agent must have made his use of others conditional upon their consent, as a matter of moral principle. As a result, although both accounts attribute a crucial role to actual consent, on the agent-focused actual-consent interpretation Kant’s prohibition turns out to be more demanding, as illustrated with the genocidal dictator example in the next section.

Some further clarifications concerning the required consent are in order. First, actual consent needs to satisfy certain normative criteria in order genuinely to count as actual consent. As mentioned, apparent consent could in fact be spurious, for example if it is the result of deceit, misinformation, misunderstanding or manipulation. Furthermore, in order to be able to give genuine consent, the person whom you want to use as means needs to be mentally competent and mature enough, and the person also needs to have and understand the relevant information. The person needs to know which end you want him to serve, how you plan to use him, what this will require of him and so on.¹⁸ Genuine agreement between two parties presupposes that they have a shared understanding of the agreement and a ‘united will’ (MM, 6: 271–2). A full specification of Kant’s account of the requirements for genuine actual consent also involves determining whether consent should always be explicit, whether consent to a general policy implies consent to its specific applications, and so on.

Second, it is important to note that those who give consent ought to meet specific moral requirements governing their own agency. Kant emphasizes that you ought not to use yourself merely as a means either (G, 4: 429, MM, 6: 420–37), and this poses constraints on the types of use to which you are morally allowed to consent (cf. MM, 6: 236). You should not consent to being used as an accomplice in a scheme of deception, for example, and you should not agree to sell yourself into slavery. Thus there is a set of normative conditions that should be met by anyone who is asked to serve as a means. Agents intending to use others as means in turn ought to take these conditions into account when asking for consent, since they must not ask others for their consent to being used in degrading or otherwise morally impermissible ways. Not all moral conditions that govern the giving of consent are negative, of course. On Kant’s view you have a duty of beneficence, so under certain circumstances you may well have a moral duty to consent. But note that even when you ought to consent, using you without your consent still counts as using you merely as a means. Kant’s Formula of Humanity does not permit, say, tricking an egoist into contributing to a good cause.

Providing a full account of Kant’s conception of genuine actual consent, as well as of the morality of giving and requesting consent, lies beyond the scope of this article.

4. Advantages of the Agent-Focused Actual-Consent Account

In the previous section, I argued for the agent-focused actual-consent interpretation of Kant’s prohibition on the basis of textual evidence. In this section, I examine whether it can avoid the difficulties associated with alternative interpretations. Let us revisit the interpretations mentioned in section 2.

4.1. Bare Actual Consent to Being Used, in a Particular Manner, as a Means to the Agent’s End

We already saw that the restriction to cases of use (rather than treatment) obviates the worry regarding the unconscious jogger in need of resuscitation: the beneficent Kantian does not use her as a means and, by implication, not merely as a means.

In the example of the genocidal dictator, the proposed agent-focused actual-consent account leads to different results than the bare actual consent account. On the proposed account, the act-utilitarian’s consent does not suffice for the dictator to avoid using him ‘merely’ as a means. When we look at the genocidal dictator’s practical reasoning, it becomes clear that she indeed uses all of her victims merely as means. Whatever maxim she is reasoning from, she patently fails to limit her use of others to cases in which they consent. The fact that the act-utilitarian happens to consent makes no difference to this assessment of the dictator’s manner of acting.

We reach the same result if we modify the example such that the dictator uses her victims’ consent as a selection criterion without treating their consent as a moral constraint. Imagine she decides to use only victims who happen to consent – say, because she perversely finds this funny. Although she restricts herself to using consenting victims, she does not ‘simultaneously use them as ends’. She acts purely on the basis of her contingent personal preference and uses others as she pleases. If she felt like it, she would instead select those who protest most loudly.

In a similar manner, the proposed account of Kant’s prohibition can handle cases of consensual enslavement and cases of consent in the context of enslavement. A slave master uses enslaved persons merely as means, regardless of whether they happen to consent to the use he makes of them. After all, were he to adopt the principle of using them only on condition that they consent, they would effectively no longer be his slaves (on Kant’s understanding of the term, MM, 6: 283). Furthermore, when two people voluntarily enter into a master–slave relationship, the prospective master must adopt, at that moment, the principle of using the slave as a means to his ends without making this use conditional upon the slave’s consent. By adopting this principle and using his slave, he clearly violates Kant’s prohibition on using others ‘merely as means’. And if a slave master allows his slaves to rest during the hottest part of the day,¹⁹ he still counts as using them merely as means, even though he gives them something they want. For he is not limiting his use of them, as a matter of moral principle, to that to which they consent. He is using them as slaves, and the decision whether to let them rest is solely up to him.

4.2. Possible Rational Consent to Being Used as Means to the Agent’s End

Unlike the possible rational consent account, the proposed account does not imply that an agent may permissibly use others as means without their actual consent, letalone over and against their explicit dissent. Their actual consent serves as a limiting condition. Thus, contrary to Parfit’s claim, the agent on the footbridge does use the heavy man merely as a means when she uses him as a trolley stopper to save five others.²⁰ She uses him as a means to her end, without his consent, so she clearly fails to respect the morally required limiting condition.²¹

4.3. Possible Consent to the Action

The fact that the proposed account covers all cases of instrumental use ensures that its construal of the criterion is not too narrow. It is not restricted to those cases in which the nature of the action makes consent impossible. It can deal with the case of organ harvesting as well as with the false promising example. It follows straightforwardly from the proposed interpretation that if a surgeon uses a healthy person’s organs to save the lives of five patients, without the person’s consent, this surgeon uses the person merely as a means.

4.4. Possible Consent to the Maxim

Finally, the fact that the account covers only cases of instrumental use ensures that the criterion is not too wide. The restriction to cases of use implies that cases such as the drowning example do not get classified as instances of the agent’s using someone merely as a means. The agent in the drowning case is not using the drowning person as a means to her ends, and by implication she is not using the drowning person merely as a means.

This result should not be misunderstood as implying that the agent is acting permissibly in letting the person drown. The prohibition against using someone ‘merely as a means’ is only one part of the Formula of Humanity; obeying it does not exhaust your moral duties. Kant claims that the formula requires that you ‘promote’ humanity as an end in itself (G, 4: 430). His first two examples in the Groundwork are cases in which the agent violates this requirement by using humanity merely as a means. The third and fourth examples are cases in which the agent does not do this but nevertheless violates the requirement of promoting humanity as an end. Even without a detailed specification of what the latter requirement entails – which lies beyond the scope of this article – it is clear from Kant’s fourth example, which involves a person who refuses to help others in need, that you violate this requirement if you refuse to help a drowning person out of self-interest or moral indifference. Doing so is morally wrong, but this is a different type of violation than that of using someone merely as a means (G, 4: 429–30). The same goes for humiliation and contempt: these do not involve the use of others merely as means, but they are nevertheless impermissible (MM, 6: 462–8).

5. A Worry about Consistency

One scenario mentioned parenthetically in section 2 requires further discussion, as it might seem to expose a problem for the proposed interpretation of Kant’s conception of using others merely as means. This is the case of the wannabe hero roaming the parks, who could not care less about promoting the ends of others and resuscitates the unconscious jogger solely in order to become famous. He would be using her as a means to his own greater glory, and he would also be using her merely as a means in resuscitating her, given that he goes ahead even though, being unconscious, she does not give consent. But if using others merely as means is morally prohibited, then it might seem to follow that it is morally impermissible for him to resuscitate the jogger.²² Alternatively, if it is morally permissible for him to resuscitate the jogger, then it might seem to follow that it is permissible to use others merely as means, at least in some cases.²³ Both results would be problematic. The first would attribute to Kant the view that resuscitating the jogger would be morally right for one agent and wrong for another on account of their different motives, which would contradict Kant’s own claim that beneficent actions are ‘in accord with duty’ even if they are not performed ‘from duty’ but from self-interest (G, 4: 398). The second would contradict the wording of Kant’s Formula of Humanity (‘never merely as a means’). Either way, the agent-focused account might seem to yield inconsistencies, and it might therefore appear to face a significant difficulty of its own. The response to this worry requires a discussion of the relation between the permissibility of the outward action and the permissibility of the agent’s underlying maxim.

Let us assume that the wannabe hero is indeed using the jogger merely as a means. He acts strictly from self-interest and fails to make his use of others conditional on their agreement. He uses the jogger merely as a means to his own greater glory. He has not adopted the maxim of beneficence and would let the unconscious die by the side of the road if it were not for the fact that he needs to resuscitate them to achieve his goal. Second, let us also assume that resuscitating the jogger is indeed ‘in accord with duty’ and at least morally permissible if not required by the Kantian duty of beneficence (which includes the duty to render assistance). The question is, then, whether these two assumptions are compatible. Can the claim that the wannabe hero is using the jogger merely as a means be made compatible with the claim that his resuscitating her is permissible?

The first thing to note is that he violates at least two Kantian duties: he violates both the perfect duty of respect and the imperfect duty of beneficence. He violates the duty of respect by using the woman, without her consent, as a means in the service of his personal project of acquiring fame (cf. MM, 6: 450, 462). It might sound odd to say that he also violates the duty of beneficence, given that he would be saving her life. Within Kant’s ethics, however, imperfect duties are duties to adopt certain maxims. By adopting the impermissible maxim of self-interest, rather than the morally obligatory maxim of beneficence, he violates the duty of beneficence (6: 388–90).

The fact that he acts on the impermissible maxim of self-interest in no way implies, however, that it is impermissible for him to resuscitate the jogger. This point can be explained with the help of Kant’s own parallel example of a shopkeeper who refrains from overcharging children, not on the basis of the morally required maxim of honesty but on the basis of the morally impermissible maxim of self-interest (G, 4: 397).²⁴ This shopkeeper is acting from self-interest rather than from duty. Nevertheless, Kant claims that his pricing is ‘in accord with duty’, since it happens to match the pricing required by the duty of honesty. Put in more colloquial terms, he does the right thing (he refrains from overcharging children), but he does it for the wrong reason (he does it on the basis of his maxim of self-interest). In sum, the fact that the shopkeeper acts on the basis of the impermissible maxim of self-interest does not make his action impermissible (charging children the regular price).

By parity of reasoning, the fact that the wannabe hero acts on the basis of the impermissible maxim of self-interest does not make his action impermissible (resuscitating the jogger). If morality demands that he resuscitate the jogger, the wannabe hero’s administering CPR is ‘in accord with duty’, even though he acts from self-interest rather than ‘from duty’. His administering CPR has the same moral status as the shopkeeper’s charging children the regular price. It is not impermissible for him to save the jogger’s life.

Thus Kant’s shopkeeper example highlights the importance of distinguishing the evaluation of an outward action from the evaluation of the agent’s underlying practical reasoning. When Kant asks whether an outward action is ‘in accord with duty’, he abstracts from the agent’s reasons for performing it. When he asks whether an agent uses someone merely as a means, by contrast, he focuses on the practical reasoning that underlies the agent’s action. The wannabe hero’s action (resuscitating the jogger) is in accord with duty and hence at least permissible. But he reasons from the maxim of self-interest and uses the jogger merely as a means towards his own greater glory, and this is morally impermissible.

The worry that the proposed account of ‘using another merely as a means’ entails that it would be morally prohibited for the wannabe hero to save the jogger – or, alternatively, that using others merely as means would sometimes be permissible – conflates the moral evaluation of the action with the moral evaluation of the agent’s maxim (and the associated practical reasoning). When, in the statement of the worry, it is assumed that resuscitating the jogger is permissible, the focus is on the outward action in abstraction from the agent’s reasoning. When it is pointed out that the wannabe hero is using the jogger merely as a means, by contrast, the focus is on his underlying maxim and his reasoning on its basis. As the structure of Kant’s shopkeeper example helps us recognize, the judgement that the wannabe hero’s action is ‘in accord with duty’ is compatible with the thesis that his underlying maxim is morally impermissible. Conversely, the fact that he uses the jogger merely as a means does not entail that it is morally impermissible for him to resuscitate the jogger.

In sum, the claim that it is morally permissible for the wannabe hero to resuscitate the jogger is indeed compatible with the claim that it is impermissible for him to use her merely as a means in doing so, and the alleged inconsistencies do not emerge.

6. Concluding Remarks

The textual evidence indicates that the prohibition on using others merely as means demands that you make the use of others as means to your ends conditional on their genuine actual consent, as a matter of moral principle. Understood in this way, the prohibition articulates a concrete moral guideline that combines a characteristically Kantian focus on the agent’s practical reasoning with the equally characteristically Kantian emphasis on the moral standing of persons as ends in themselves. Moreover, this account of the prohibition avoids the difficulties associated with alternative interpretations.

As a final comment, let me note that this account also fits well with widely shared moral convictions about the importance of informed consent, convictions that are generally believed to have Kantian provenance. ‘Informed consent’ requirements, for example in medical research involving human subjects, are generally understood as imposing an obligation on the researcher. The researcher ought to use human test subjects only on condition that they genuinely consent to her using them in the proposed manner and for the stated end.²⁵ Whether she properly discharges this duty depends not on whether her test subjects just happen to consent. Rather, it depends on whether she makes their participation in the study conditional upon their genuine actual consent, by following procedures designed to ensure that potential test subjects receive and understand the necessary information and participate voluntarily. This common understanding of informed consent requirements fully matches Kant’s prohibition as understood on the agent-focused actual-consent account developed in this article; it matches the prohibition only partially, or not at all, when this prohibition is interpreted along the lines of the alternative accounts discussed above. Hence the account offered here also closes the current gap between the scholarly interpretation of Kant’s prohibition and important ethical codes that are believed to be inspired by it.²⁶