Why Humean constructivists should become Kantian constructivists

Abstract

One of the main reasons for philosophers to have embraced Humean constructivism rather than Kantian constructivism is a negative one: they believe that in the end Kantian constructivism is an unstable position. Their idea is that the Kantian constructivist can either choose to hold on to the idea of categorical reasons for action but in that case she has to be prepared to commit to (robust) moral realism (which both Humean and Kantian constructivists reject) or alternatively, she might reject (robust) moral realism but in that case she has to give up on the idea of categoricity. The aim of this paper is to defend Kantian constructivism against Humean constructivism and more specifically against recent objections raised by Sharon Street. I will do so by arguing that Kantian constructivism follows from formal, normative commitments that pertain to instrumental reasoning that Humean constructivists like Sharon Street themselves accept. More specifically I will argue that categorical reasons for action follow from applying the principle of instrumental rationality to the first-person perspective of an agent, provided that there are certain necessary means for action in general. From this follows, I will argue, that Humean constructivists should either become Kantian constructivists or that they have to become sceptics about normativity.

Keywords:

• Kantian constructivism
• Humean constructivism
• categorical reasons
• instrumental reasoning

1. Introduction

Kantian constructivism is positioned “in the space between realist and relativist accounts of ethics” (O’Neill 1990, 206). Contrary to moral realism, Kantian constructivism brackets off the existence of “mind-independent” moral facts. Contrary to moral relativism, Kantian constructivism holds that there are certain moral judgements that every agent necessarily has to accept independent of his or her contingent desires, aims or beliefs. Kantian constructivism thus aims to vindicate moral objectivity, in terms of categoricity and universality, while avoiding the well-known metaphysical, epistemological and motivational problems of moral realism (Korsgaard 2008).

Both (robust) moral realists and Humean constructivists, however, claim that Kantian constructivism is an unstable position. They argue that the Kantian constructivist can either holds on to the idea of categorical reasons but in that case she has to be prepared to commit to (robust) moral realism (see e.g. Enoch 2006, 2011; Guyer 2013), or alternatively she might rejects (robust) moral realism but in that case she has to give up on the idea of categoricity (Street 2012). The aim of this paper is to defend Kantian constructivism against Humean constructivism specifically. I will do so by replying to Sharon Street’s (2012) recent articulation of the above-mentioned dilemma for Kantian constructivism. The reason for focussing on Street is that she has given one of the most recent and influential Humean inspired criticisms of Kantian constructivism and because her objection has not yet been responded to by Kantian constructivists.

I will respond to Street’s objections by arguing that the source of Kantian, moral objectivity should be located in a special type of instrumental reasoning from the first-person perspective of an agent, which I will call categorical instrumental reasoning. Moreover, I will argue that the notion of categorical instrumental reasoning presents Street with a dilemma. The first horn of the dilemma is for the Humean constructivist to reject the normativity of the principle of instrumental rationality, but this rejection backfires insofar as Street’s Humean constructivist account of normativity also presupposes the normativity of the instrumental principle. Alternatively, the Humean constructivist could accept the instrumental principle, but then she no longer has any reason not to accept Kantian constructivism. In other words, I will try to show that Humean constructivists should either become Kantian constructivists, or become sceptics about normativity.

Before proceeding, one point of clarification is in order. Street raises her objections to Kantian constructivism in the context of Christine Korsgaard’s (1996) argument for the value of humanity in The Sources of Normativity. Street, however, explicitly stresses that her objections to Korsgaard are meant to apply to Kantian constructivism in general (2012, 45). I will therefore discuss Street’s objections as objections against Kantian constructivism in general, and not as arguments about the details of Korsgaard’s argument. The goal of this paper is thus not to defend Korsgaard’s specific variant of Kantian constructivism, but to explore whether there is a variant of Kantian constructivism that is not susceptible to Street’s objections.

The structure of the paper is as follows. In section 2, I introduce the structure of a Kantian constructivist argument for objective moral judgements in the form of a transcendental argument. In section 3, I introduce Street’s objection that Kantian constructivism cannot justify any categorical normative reasons and I briefly discuss how Street’s objection differs from David Enoch’s much discussed “shmagency objection” to Kantian constructivism. In section 4, I respond to Street’s objection by arguing that the source of Kantian objectivity should be located in categorical instrumental reasoning. In section 5, I will sketch two argumentative strategies that could be used to justify substantive categorical reasons for action on the basis of this approach.

2. A transcendental argument for objective moral judgements

Before turning to Street’s objections to Kantian constructivism, let me first briefly introduce the structure of a Kantian constructivist argument for objective moral judgements. By way of example, I will discuss Korsgaard’s argument for the value of humanity as an objective moral value.¹ The reason for focusing on Korsgaard is that even though Street puts forward general objections to Kantian constructivism, she formulates her objections in the context of Korsgaard’s Kantian constructivism. I will therefore follow Street by using Korsgaard’s Kantian constructivism as an example to introduce the general structure of Kantian constructivist argument for objective moral judgements.

Korsgaard summarizes her argument as follows:

Since you are human you must take something to be normative, that is, some conception of practical identity must be normative for you. If you had no normative conception of your identity, you could have no reasons for action, and because your consciousness is reflective, you could then not act at all. Since you cannot act without reasons and your humanity is the source of your reasons, you must value your humanity if you are to act at all (1996, 123).

This argument raises many questions: what does Korsgaard mean by the value of humanity? Why can only humanity function as the source of reasons? Why must you take something to be normative for you? Although I think these questions are essential if we want to assess (or indeed understand) Korsgaard’s argument, I will not try to reconstruct her argument here.² Instead, I want to focus on the type of argument Korsgaard makes rather than on the details of her argument.

Korsgaard writes: “the argument I have just given is a transcendental argument” (1996, 123). A transcendental argument is an argument which starts from a certain inescapable aspect of our self-understanding, for instance our self-understanding as experiencers, thinkers or agents, and consequently raises the question of what makes this self-understanding possible. More specifically, a transcendental argument is an argument which aims to show that a possibly non-obvious and non-trivial Y is a necessary condition for the possibility of X – where, given that X is somehow inescapable for A, it logically follows that A must accept Y:

1. X is inescapable for A

2. Y is a necessary condition for the possibility of X

3. Therefore, A must accept Y

Transcendental arguments are employed both in theoretical philosophy and in practical philosophy. In the remainder of this paper, however, I focus exclusively on their use in practical philosophy.³

Korsgaard’s transcendental argument can be roughly summarized as follows:

1. Agency is inescapable for A (i.e. you, me or any other human being)

2. Valuing your humanity is a necessary condition for the possibility of agency

3. Therefore, A must value his/her (own) humanity

The most fundamental feature of this kind of argument is that it proceeds from the first-person perspective. A transcendental argument is not a type of conceptual analysis, but it tries to explore the necessary commitments of beings who understand themselves in a specific way, e.g. beings who understand themselves as agents. Consider, for instance, the first premise of a transcendental argument, i.e. the claim that X is inescapable. Korsgaard’s argument for the inescapabilty of acting for reasons proceeds as follows:

I perceive, and I find myself with a powerful impulse to believe. But I back up and bring that impulse into view and then I have a certain distance. Now the impulse doesn’t dominate me and now I have a problem. Shall I believe? Is this perception really a reason to believe? I desire and I find myself with a powerful impulse to act. But I back up and bring that impulse into view and then I have a certain distance. Now the impulse doesn’t dominate me and now I have a problem. Shall I act? Is this desire really a reason to act? The reflective mind cannot settle for perception and desire, not just as such. It needs a reason. Otherwise, at least as long as it reflects, it cannot commit itself or go forward (Korsgaard 1996, 93).

Korsgaard claims that self-conscious beings are presented with the question of whether a desire or inclination is really a reason to act and that they therefore need a reason for action. This is a claim about the first-person perspective of an agent. From a third-person perspective, we might understand ourselves as governed by causal laws. But this is not how we can understand ourselves from the first-person perspective.⁴ This first-person perspective is one of the essential characteristics of a transcendental argument and, as I will show below, neglecting this feature lies at the basis of the Street’s objections to Kantian constructivism.

Before proceeding, one further point of clarification is in order. If Korgaard’s, or a similar, transcendental argument is successful it “shows only (or at most) that you must place a value on your own humanity, but not yet that you therefore have obligations to other human beings” (Korsgaard 1996, 130). In other words, if you must accept whatever are the necessary preconditions for your agency, this does not imply anything about interpersonal morality, e.g. it does imply that you must also value the humanity of other agents. An additional argument is needed to arrive at claims of interpersonal morality.⁵ Although showing how interpersonal morality can be justified through a transcendental argument is essential in defending a comprehensive account of Kantian constructivism, this paper only discusses the first part of the argument, i.e. the argument for categorical reasons. The reason for focusing on the first part of the argument is that Street herself explicitly limits her discussion to the first part of the argument (2012, 48f15).

3. The no-normativity objection

A transcendental argument aims to show that any agent necessarily has to accept whatever are the necessary conditions of agency, and that these necessary conditions of agency include categorical claims or reason. Street’s no-normativity objection states that the claim “if you are an agent, you must Y” – e.g. Korsgaard’s claim “if you are an agent, you must value your humanity” – is ambiguous and ultimately incoherent (Street 2012, 51).⁶ In this section, I will respond to this objection and in this way I will try to clarify the way in which a transcendental argument can lead to categorical normative claims.

In order to argue for the conclusion that a transcendental argument cannot lead to any categorical normative claim, Street distinguishes two possible interpretations of the statement “if you are an agent, you must Y,” and she argues that both fail to justify a categorical normative claim.

On the first, normative interpretation, such claims “state substantive normative truths” (Street 2012, 51), meaning that there is a mind-independent reason to value your humanity. Thus, according to the normative interpretation, the claim “if you are an agent, you must Y” should be read as “if you are an agent, you have a mind-independent, substantive reason to Y”. The problem with this interpretation should be obvious. Insofar as constructivism aims to provide an alternative for moral realism, it should not presuppose any substantive normative truths in its argument for objective moral claims. Street therefore concludes that the normative interpretation is unavailable for constructivists.

On a second, conceptual interpretation, on the other hand, “these claims state conceptual, ‘constitutive’ truths, such as ‘To be a parent, one needs to have children’” (Street 2012, 51). If the claim “if you are an agent, you must Y” is read as a conceptual claim, the argument clearly does not presuppose any substantive, mind-independent reason. In this sense, the conceptual claim is available to Kantian constructivists (whereas the normative interpretation is not). The problem with the conceptual interpretation, Street argues, is, however, that it is merely conceptual, i.e. it does not generate any reasons for action. Street writes that “an ‘agent’ who doesn’t take anything at all to be a reason does not have a reason to take something or other to be a reason; rather, ‘he’ or ‘she’ is not an agent at all” (Street 2012, 51). Street’s objection is thus that the transcendental claim either presupposes moral realism because it relies on a “substantive normative truth” or does not generate any reason for action because it is merely a conceptual claim.

Street’s no-normativity objection to Kantian constructivism should be distinguished from David Enoch’s (2006, 2011) “shmagency objection” to so-called constitutivist theories of (moral) normativity, of which Kantian constructivism is a subclass. Enoch imagines a sceptic about Kantian constructivism, what he calls “a shmagent”, as saying: “Classify my bodily movements and indeed me as you like. Perhaps I cannot be classified as an agent without aiming to constitute myself [or without doing whatever is a necessary condition for the possibility of agency]. But why should I be an agent?” A shmagent is thus someone who accepts that agency has certain necessary conditions of possibility, but who denies that these generate any normative reasons, because he or she simply does not have a reason to be an agent.

Enoch’s objection to Kantian constructivism (and constitutivism in general) is that whatever are the necessary preconditions of agency, these can only generate reasons for action insofar as one has a reason to be an agent in the first place. Enoch grants that agency is inescapable, which implies that the shmagent is also an agent⁷ and that there might be necessary preconditions of agency (2006, 177–178). His objection is that these necessary conditions of agency can only be normative for agents if agency itself is normatively inescapable, i.e. if there is a reason to be an agency. In other words, Enoch believes that a transcendental argument in ethics can only work if we introduce an additional premise (1* below):

1. X is inescapable

1*. X is also normatively inescapable, i.e. one necessarily has reason to X

2. Y is a necessary condition for the possibility of X

3. Therefore, you must Y

Enoch claims that making explicit premise 1* shows that Kantian constructivism cannot be a fundamental theory of normativity, because it has to rely on a mind-independent reason to get off the ground (2006, 192).

Although Street also questions the normativity of Kantian constructivism, her objection is directed at a different part of the Kantian constructivist argument. Street does not think that one needs a reason to be an agent in order to generate normativity from the standpoint of agency – in fact, her own Humean constructivism also presupposes that normative reasons can be generated from the perspective of agents (I will come back to this below). Street’s objection is directed at premise 2 of the transcendental argument that “Y is a necessary condition for the possibility of X”. More specifically, she denies that necessary conditions for the possibility of agency express a normative relation.

4. Categorical instrumental reasons

Street’s dilemma regarding Kantian constructivism is that either it cannot justify normative reasons or it relies on a mind-independent normative reason. However, this supposed dilemma overlooks the possibility that there might a third way to interpret the claim that “if you are an agent, you must Y”. There might be normative reasons that are not derived from mind-independent reasons, but which are generated from the first-person perspective of the agent.

In this section, I will elaborate one possible way in which Kantian constructivism could argue for this conclusion. The proposal is to understand the necessary conditions for the possibility of agency in terms of categorical instrumental reason. I will argue that although most instrumental reasons are hypothetical, insofar as they depend on taking a contingent end, there are also instrumental reasons involved in adopting any end whatsoever. These are what I will call categorical instrumental reasons. Note that this reply crucially hinges on a constructivist justification of the principle of instrumental rationality. I will briefly come back to the justification of the instrumental principle towards the end of the section. The goal of his section is, however, limited. I aim to show that Street, or any other philosopher with Humean constructivist sympathies, has to accept the coherency of the Kantian constructivist project, given their own philosophical commitments, which includes, as I will argue below, the acceptance of the principle of instrumental rationality.

The account of Kantian constructivism presented in this section is very much in line with Alan Gewirth’ (1978) interpretation of transcendental arguments. It is questionable whether Korsgaard would or could agree with this account of Kantian constructivism.⁸ In this section I am, however, not so much concerned with this exegetical question, but with trying to provide one possible Kantian constructivist response to Street’s objection.

Consider, first, contingent instrumental reasons. The principle of instrumental reasoning states that if I want to achieve an end, I have a reason to take the necessary means to that end or give up the end. For instance, if I want a drink, and there is only water available, I have a reason to drink water. Of course, once I discover that there is only water available, I might revise my end and decide that I would rather be thirsty than drink water, but insofar as I want a drink, I ought to drink the water.

Contingent instrumental reasons are obviously contingent on pursuing a specific end. Only someone who wants a specific end has reason to take the necessary means to that end, and once someone discovers what the necessary means to that end are, she might decide to revise the end, instead of taking the necessary means to that end. So although contingent instrumental reasons are compatible with constructivism, insofar as instrumental reasons do not presuppose substantive normative truths but only agents who pursue ends, it is not sufficient to argue for constructivist, objective reasons.

In order to argue for Kantian constructivism, an additional step is required. The additional step is the recognition that there might not only be instrumental reasons that are conditional on the acceptance of a contingent end but that there might also be instrumental reasons involved in adopting any end whatsoever. For instance, being alive is a necessary means for engaging in any action whatsoever. Independent of whether one wants a drink, wants to play chess, study philosophy or start a family, etc., a necessary means for acting, whatever one’s contingent ends, is that one is alive. Even someone who commits suicide has to be alive in order to achieve his or her end (cf. Gewirth 1978, 136–37). In this sense, one has reason to want to be alive, independent of one’s contingent ends.⁹

Insofar as one has a reason to take the necessary means to achieve an end, and insofar as there are necessary means for engaging in any action whatsoever, there are categorical reasons to take the necessary means for engaging in action as such. The only difference between contingent and categorical instrumental reasoning is that whereas contingent instrumental reasons only exist insofar as someone wants a specific end, categorical instrumental reasons exist insofar as there are beings who want to achieve any end whatsoever. The categorical instrumental means are categorical insofar as they do not depend on one’s contingent desires or preferences, but only on one’s agency as such. Contrary to Street (2012, 53) and many others, to say that a reason is categorical therefore does not have to mean that a reason is purely non-instrumental. It could be, but it need not be. Instead, saying that a reason is categorical means that it should be accepted independent of one’s contingent preferences or desires, not that it should be accepted independent of being an agent.

By forcing the Kantian constructivist in the dilemma to choose between relying on substantive normative truths or expressing merely conceptual truths, Street neglects the crucial fact that Kantian constructivists try to justify objective reasons from the perspective of the self-understanding of agents. Korsgaard, for instance, describes the importance of the first-person perspective as follows:

Value, like freedom, is only directly accessible from within the standpoint of reflective consciousness. And I am now talking about it externally, for I am describing the nature of the consciousness that gives rise to the perception of value. From this external, third-person perspective, all we can say is that when we are in the first-person perspective we find ourselves to be valuable, rather than simply that we are valuable. There is nothing surprising in this. Trying to actually see the value of humanity from the third-person perspective is like trying to see the colours someone sees by cracking open his skull. From outside, all we can say is why he sees them (Korsgaard 1996, 124).

In other words, it would be a mistake to think that the claim “if you are an agent, you must Y” expresses merely a conceptual truth. This would be to take an external, third-person perspective on instrumental reasoning. The claim “if you are an agent, you must Y” is a claim about what any agent is necessarily committed to from her first-person perspective as an agent, i.e. as someone who wants to achieve ends (whatever the contingent content of those ends).

Note that when I say that the transcendental claim (“you must Y”) is not merely a conceptual truth, I do not want to deny that the instrumental principle might also be a conceptual truth. What is important, however, is to notice that even if the instrumental principle is a conceptual truth, it does generate normativity if it is applied to the first-person perspective of an agent. From the first-person perspective of someone who tries to achieve certain purposes, the instrumental principle generates normativity generates normativity because it claims that one ought to take the necessary means to one’s goals. This is what I mean when I say that a transcendental must is not merely a conceptual truth – this contrasts with the claim that “to be a parent one needs children”, which is merely a conceptual truth, because it does not generate any reasons for action even when understood from the first-person perspective of an agent.

In my response to Street, I have relied on the normativity of the instrumental principle. One might legitimately raise the question of the source of the normativity of the instrumental principle. I will briefly come back to this question at the end of this section, but let me stress that, as I already stresses at the beginning of this section, responding to Street’s specific objection does not require arguing for the instrumental principle. The reason for this is that Street herself actually accepts the normativity of the principle of instrumental reasoning, and she believes that it can be given a constructivist foundation (2008, 228). Consider, for instance, her remark that according to her Humean constructivism, “the agent might of course still be making an instrumental mistake as dictated by the terms of his other values” (Street 2012, 55f23).¹⁰ In other words, Street’s objection does not question the normativity of the instrumental principle, and because she herself relies on the normativity of the instrumental principle it seems legitimate to accept its normativity in the context of discussing her objection to Kantian constructivism.

This has a surprising implication. For if I am correct in believing that a transcendental argument should be understood as generating categorical instrumental reasons, this actually presents Street, and not the Kantian constructivist, with a dilemma. One option is for Street to reject the normativity of the instrumental principle, but this rejection backfires insofar as Street’s Humean constructivism also presupposes the normativity of the instrumental principle. Rejecting the instrumental principle therefore not only undermines Kantian constructivism but also Humean constructivism, which seems to imply that denying the instrumental principle might lead to scepticism about normativity (at least assuming, as Street also does, that moral realism is off the table as well).¹¹

Alternatively, Street could accept the instrumental principle but then she would have to accept that Kantian constructivism is a coherent project, because the normativity of categorical instrumental reasoning is not more mysterious than the normativity of contingent instrumental reasoning. Opting for the second horn of the dilemma, however, similarly undermines Street’s Humean constructivism. After all, Street’s argument for Humean constructivism is essentially premised on the rejection of Kantian constructivism. Street writes:

[Humean constructivism] is defined negatively in contrast to Kantian constructivism: it is characterized by a general skepticism about arguments which seek to establish moral values as following from the standpoint of “agency” as such. Therefore, the obvious way, and perhaps the only way, to argue for Humean as opposed to Kantian constructivism is negatively – by examining the best attempts to establish such substantive values and showing why they fail (Street 2012, 45).

Accepting that Kantian constructivism is a coherent position therefore not only implies that Street’s objections against Kantian constructivism fail but that her main argument in favour of Humean constructivism fails as well. In short: Street has to choose between scepticism about normativity or Kantian constructivism. It should be clear which horn of the dilemma is to be preferred.

But what does a Kantian constructivist have to say to a sceptic about instrumental normativity? The dominant Kantian constructivist justification of the instrumental principle is to argue that accepting the instrumental principle is itself constitutive of pursuing a purpose or having an end, i.e. that is constitutive of agency.¹² James Skidmore, for instance, claims that an agent who denies the instrumental principle would be unintelligible (Skidmore 2002, 138). Skidmore illustrates this point by imagining a conversation between a “questioner” and a “sceptic” about the instrumental principle. The conversation takes place on a Friday afternoon, and the sceptic just told the questioner that he is planning to go the beach for the weekend. Skidmore images that the following conversation unfolds after the sceptic told the questioner about his plans for the weekend:

Questioner

Sounds nice. How are you going to get there on such short notice?

Skeptic

I don’t know. What difference does that make?

Questioner

You know, you might be able to catch the bus there. I think the last one leaves in an hour.

Skeptic

Yes, I know. But I hate riding the bus.

Questioner

So you have some other way to get there?

Skeptic

No.

Questioner

But you said you intended to spend the weekend at the ocean.

Skeptic

I do. They say the weather will be beautiful.

Questioner

And the only way you can get there is by taking the bus?

Skeptic

Yes, as far as I know, but what’s your point? I’m not taking the bus. Ah, but how wonderful it will be to spend a couple of days relaxing on the beach … (Skidmore 2002, 129)

Skidmore’s point is that the sceptic becomes unintelligible as an agent, i.e. as someone who pursues an end, and that the sceptic therefore has to accept the instrumental principle.

Although much more needs to be said about the justification of the instrumental principle, I will assume that any purposive agent should indeed accept the instrumental principle. After all, worries about Kantian constructivism typically focus on their ambition to justify categorical and universal claims about interpersonal morality and not their ability to justify instrumental normativity.¹³

5. Substantive categorical instrumental reason

In the previous section, I have argued that Kantian constructivist normativity is not more mysterious than, for instance, Humean normativity.¹⁴ The transcendental conditions that can lead to normative reasons are those claims that are instrumentally necessary to be an agent. However, even if my argument succeeds, I would have merely shown that Kantian constructivism is a coherent project.

One might therefore object that I have merely shown that there is conceptual room for the Kantian constructivist to respond to Street’s objections. The real worry, however, is not just whether or not Kantian constructivism is a coherent project, but whether or not Kantian constructivism can justify any substantive categorical reasons. In other writings, Street, for instance, states that she “den[ies] that the rabbit of substantive reasons can be pulled out of a formalist hat” (2010, 370).¹⁵ If there are no substantive moral reasons that can be justified by a Kantian constructivist, the conclusion that Kantian constructivism is a coherent project might be nothing more than a Pyrrhic victory.

This kind of “emptiness objection” to Kantian ethics is a recurring objection to Kantian ethics and much has been written on the alleged emptiness of Kantian ethics.¹⁶ Fully addressing this worry therefore lies beyond the scope of this paper. But let me, however, indicate how the idea of categorical instrumental reason provides the tools to respond to this objection.

One reason for thinking that the account sketched in this paper can lead to substantive reasons is that the principle of (categorical) instrumental reason is not just a formal principle.¹⁷ The material content of this principle is provided the necessary means of action. Or to be more precise: the principle of instrumental reasoning itself is a formal principle, but it has material content once applied to the first-person perspective of an agent. Consider, first, a case of contingent instrumental reason, e.g. the case of me wanting a drink. If we apply the instrumental principle to the first-person perspective of someone who wants a drink, it generates a substantive reason to drink water. The principle of contingent instrumental reason therefore generates substantive reasons as long as there are beings that pursue ends.

Categorical instrumental reasoning works in exactly the same way: substantive reasons are generated from the perspective of someone who wants an end. However, as opposed to contingent instrumental reasoning, categorical instrumental reasoning focuses on the means that are necessary to engage in any action whatsoever. The specific challenge for categorical instrumental reasoning is thus to identify substantive means that are necessary for engaging in any action whatsoever.¹⁸

In section 4, I have already mentioned one example: being alive seems to be a necessary means to engage in any action whatsoever. That is, it is impossible to pursue any end whatsoever if one is not alive. One therefore has a substantive reason to want to be alive insofar as one engages in any action whatsoever, even if one wants to commit suicide. This is, however, but one example of a necessary means of agency and it arguably leads to trivial results because, in the case of someone who wants to commit suicide, it only generates a reason to stay alive just long enough to commit suicide.¹⁹

There are two ways to move beyond this trivial result. The first is to identity further necessary means of agency beyond subsistence. This is a complex exercise which involves both a further specification of what it means to be an agent and a consideration of relevant, general anthropological insights concerning the abilities and vulnerabilities of human agents. For instance, it seems clear enough that one needs a certain amount of freedom in order to pursue any end whatsoever. But how much freedom one needs depends on how exactly one conceptualizes both agency and freedom. For instance, on a broadly republican account of freedom, one might want to argue that a civic condition is a necessary condition for the possibility of acting for any end whatsoever, because the civic condition is the only condition in which one is not susceptible to being subject to the will of another person. But independent from the details of these further necessary means of agency, it seems likely that they will include more than just subsistence.

The second way to move beyond this trivial result is to acknowledge that the argument discussed in this paper is only one part of a Kantian constructivist argument for moral objectivity. As already mentioned at the end of section 2, Kantian constructivists put forward an additional argument for the claim that one has moral obligations towards (all) other persons. Street does not discuss this second part of the Kantian constructivist argument because she believes the argument discussed above already fails. However, if my argument for the claim that one has a substantive reason to want to be alive insofar as one engages in any action whatsoever is combined with an additional argument for interpersonal morality, it would show that any agent has a categorical reason to respect or protect the life of all other agents. And although a reason to be alive might be trivial in the intrapersonal case (as discussed), it is far from trivial in the interpersonal case.

6. Concluding remarks

In this paper, I have put forward a defence of Kantian constructivism against a recent Humean inspired objection raised by Sharon Street. I have argued that the source of normativity of Kantian constructivism should be found in categorical instrumental reasoning from the first-person perspective on an agent and I have tried to show how this approach could be used to justify substantive categorical reasons for action.

Let me end by stressing that although I think Street’s objections to Kantian constructivism fail, I think she is completely right to put pressure on Kantian constructivists to clarify their methodology. This paper is an attempt at such a clarification. Although the proof of the Kantian pudding is ultimately in the eating, I hope at least to have shown that Kantian constructivism is a coherent and promising project. In fact, given the well-known problem of moral realism, there are good reasons to take Kantian constructivism more seriously than has been done so far.²⁰

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Sem de Maagt is an assistant professor in ethics and political philosophy at Utrecht University, The Netherlands.

Notes

1 I will not discuss the relation between The Sources and Self-Constitution (2009) in this paper. The reason for this is that the goal of this paper is not to defend Korsgaard’s specific conception of Kantian constructivism, but Kantian constructivism in general. Note also that, in Self-Constitution, Korsgaard claims: “I am always making the same argument” (2009, 76).

2 For a discussion of different ways to reconstruct Korsgaard’s argument see Stern (2011).

3 Discussions about transcendental arguments in theoretical philosophy mainly revolve around the question of whether the conclusion of a transcendental argument is a justified true belief (strong transcendental argument) or “merely” a justified belief (modest transcendental argument). The locus classicus is that of Stroud (1968), who argues that transcendental arguments cannot justify any ontological claim. For a recent overview of transcendental arguments in theoretical philosophy see Stern (2000, 2013). In this paper, I focus exclusively on transcendental arguments in ethics and I will not discuss their relation to transcendental arguments in theoretical philosophy. Let me, however, note two important differences between transcendental arguments in these two domains. First, whereas transcendental arguments in theoretical philosophy often try to justify an ontological claim, transcendental arguments in practical philosophy typically do not have this ambition, at least insofar as they are employed by constructivists who either reject or bracket the existence of moral facts (an exception is Enoch [2011, 50–84]). Transcendental arguments in ethics therefore might not be susceptible to Stroudian worries (cf. Robert Stern 2013). Second, whereas the explanandum of a transcendental argument in theoretical philosophy is a justified (true) belief (e.g. belief in our existence), the explanandum in practical philosophy is a normative judgement, i.e. a judgement about what one should do (e.g. value one’s humanity).

4 For a discussion of the difference between a third-person and a first-person perspective on agents see Korsgaard (1989, 1995).

5 Korsgaard, for instance, puts forward an argument about the publicity of reasons (1996, 131–166). Alan Gewirth, on the other hand, puts forward “the argument from the sufficiency of agency” which states, roughly, that given that agency is a sufficient condition for being committed to valuing the necessary means of agency, one has to value not only the necessary means of one’s own agency but also the necessary means of every agent (on pain of denying that agency is a sufficient condition for valuing the necessary means of agency) (Gewirth 1978, 101). For a critical comparison of different Kantian constructivists arguments for interpersonal morality see Beyleveld (2015) and de Maagt (2018, forthcoming).

6 Michael Nance (2015) puts forward an almost identical objection to Fichte’s transcendental argument for the necessity of mutual recognition. He presents the objection in terms of what he calls a “modal dillemma”:

The modal dilemma goes as follows. Fichte understands it to follow from his argument that “I must in all cases recognize the other as a free being, i.e., I must limit my freedom through the concept of the possibility of his freedom” (ibid.: 39). There is an ambiguity concerning the nature of the “must” in this conclusion. The necessity expressed by the “must” is either metaphysical or normative. If the necessity is metaphysical, then Fichte’s conclusion is falsified by empirical reality, since one can evidently be a finite self-positing agent without recognizing the rights of others. […] But if the necessity expressed by the “must” is normative – if his conclusion means “I ought in all cases to recognize the other” – then it does not follow from the premises about self-positing that Fichte establishes in the preceding theorems of his deduction (Nance 2015, 615–16).

This seems to suggest that Street’s objection to Korsgaard’s argument is an instantiation of a more general objection to transcendental arguments in ethics. I will not discuss the applicability of this no-normativity or modal objection to Fichte. Instead I will try to show that there is an interpretation of transcendental arguments in ethics which is not susceptible to this kind of objection.

7 “‘Perhaps,’ Korsgaard’s sceptic may say, ‘I cannot opt out of the game of agency, but I can certainly play it half-heartedly, indeed under protest, without accepting the aims purportedly constitutive of it as mine.’ The kind of necessity the game of agency has to enjoy in order to solve the problem we are now in is normative necessity. Invoking other necessities here will just not do” (Enoch 2006, 188; see also 2011, 216).

8 Korsgaard argues that “the instrumental principle cannot stand alone” (1997, 251). Instead, Korsgaard claims that the instrumental principle can only generate normativity if there are already ends that are good. This conception of the role of the instrumental principle is incompatible with the response in this section, because my response claims that (categorical) instrumentality can generate normativity without presupposing anything about the goodness of ends themselves. At the same time, however, Korsgaard also believes that the distinction between the instrumental principle and the categorical imperative might “break down” because “moral or unconditional principles and the instrumental principles are both expressions of the basic requirement of giving oneself a law, and bring out different implications of that requirement” (1997, 250f73). This seems to bring her view closer to my view.

9 The point of this paragraph and the example of a categorical reason to want to be alive is merely to illustrate the formal structure of the Kantian constructivist argument. I will come back to the question of the substance of categorical reasons in the next section.

10 In fact, in other writings, Street acknowledges that the instrumental principle can generate reasons for action while at the same time she claims that the instrumental principle is a conceptual truth (see Street 2008, 228f37).

11 Another option for Street would be to retreat to what Bernard Williams (1979) calls the “sub-Humean model” (1979, 102), which states that any desire also constitutes a reason. This, however, also seems to amount to scepticism about normativity insofar as on the sub-Humean model one cannot make a mistake about one’s reasons.

12 See e.g. “willing an end just is committing yourself to realizing the end. Wiling an end, in other words, is an essentially first-personal and normative act. To will an end is to give oneself a law, hence, to govern oneself […] Willing an end is equivalent to committing yourself, first-personally, to taking the means to that end […] What is constitutive of willing the end is not the outward act of actually taking the means but rather the inward, volitional act of prescribing the end along with the means it requires to yourself” (Korsgaard 1997, 245). See also (Skidmore 2002; Beyleveld and Bos 2009).

13 Acknowledging that it is necessary to accept the instrumental principle insofar as one understands oneself as an agent makes clear that although one has a prudential reason for accepting categorical instrumental reasons because they are instrumental to the pursued of your goals (whatever they may be), the judgement that one categorically instrumentally ought to take the necessary means to action is itself not prudential or instrumental but necessary insofar as one understands oneself as agent (cf. Beyleveld 2015, 594). Denying that one has to take the necessary means to action is thus not only prudentially or instrumentally irrational, but it also contradicts with the idea that one understands oneself as an agent, insofar as accepting the principle of instrumental rationality is constitutive of agency. This is another way illustrating that Kantian (objective) normative reasons are both instrumental and categorical.

14 Another way to make the same point is to understand transcendental commitments as limiting cases of internal reasons (cf. Williams 1995, 220f3). An internal reason is generally understood as a reason which can be derived, via a valid deliberative route, from the existing motivational set of a person, i.e. the existing beliefs, desires and preferences of a person (Williams 1979). A transcendental argument is a limiting case of such an internal reason, because although it does not presuppose the existence of any mind-independent reason (contrary to the external reason theories), reasons are not derived from the substantive content of the person’s motivational set (as Humeans, including Street, typically assume), but from the conditions of possibility of having a motivational set at all, independent from its contingent content.

15 Note that there is a certain tension between Street’s emptiness objection and her objection that Kantian constructivism is incoherent. This tension is due to the fact that it seems to me that it only makes sense to say that a theory is empty once it is acknowledged that the theory is at least coherent. Why bother about the emptiness of an incoherent theory? In addition, although Street repeatedly stresses that Kantian constructivism is empty, she does not give an argument for this conclusion. I do not want to try to shift the burden of proof – I do think that the burden of proof is on the Kantian constructivist to show that substantive reasons can be justified. Instead, I merely want to stress that, as it stands, this not a real objection to Kantian constructivism but merely describes an important challenge.

16 This objection to Kantian ethics goes back as far Hegel’s empty objection to the formula of universal of the categorical imperative. Hegel claims that, on the basis of the formula of universal law

no immanent theory of duties is possible. One may indeed bring in material from outside and thereby arrive at particular duties, but it is impossible to make the transition to the determination of particular duties from the above determination of duty as absence of contradiction, as formal correspondence with itself, which is no different from the specification of abstract indeterminacy; and even if such a particular content for action is taken into consideration, there is no criterion within that principle for deciding whether or not this content is a duty. On the contrary, it is possible to justify any wrong or immoral mode of action by this means. (1991, 162;135).

For recent overviews of the discussion of the emptiness objection see (Freyenhagen 2012, 45–46; Stern 2012, 75).

17 For a discussion of why formal principles are not necessarily empty principles see Herman (1996, 217).

18 This project is similar to Alan Gewirth’s (1978) attempt at defining certain “generic features of action,” which he defines as “the invariant features that pertain generically to all actions” (1978, 25). Compare also John Rawls concept of “primary goods,” which are supposed to be “these goods [that] normally have a use whatever a person’s rational plan of life.” (1999, 54).

19 I am thankful to an anonymous reviewer for raising this objection.

20 An earlier version of this paper was presented at the “Objectivity in Ethics” conference at Utrecht University (2016). I am grateful for the comments and questions I received on this occasion. Special thanks to Rutger Claassen, Marcus Düwell, Fleur Jongepier, Wouter Kalf, Michael Klenk and Ingrid Robeyns for their helpful comments on various earlier drafts of this paper. I would also like to thank an anonymous reviewer for his or her constructive comments and suggestions.