2.1

As I use the term, a capacity is a potentiality to do something, ranging from the nutritional activities of plants and animals to the conscious and self-conscious activities of animals. It does not include everything in the life of an organism. Whereas walking is an activity, being burned to a crisp is not. The potentiality to walk is a capacity of mine. The potentiality to burn is not. Exercises of capacities are thus doings of organisms.

One capacity differs from another given what they are potentialities to do. My capacity to confide differs from my capacity to keep a secret, say, given how confiding differs from keeping a secret. An account of the nature of a capacity thus mirrors an account of an activity. As we might put it, the principle that describes the nature of a capacity spells out what it is a potentiality to do. And that principle is a standard for its exercise. If I exercise my capacity to confide with respect to my crush about my feelings for him, to convey is correct, to conceal incorrect. If I instead exercise my capacity to keep a secret, to conceal is correct, to convey incorrect. Which capacity I exercise thus determines the standard operative.

Exercises of capacities can be correct or incorrect because an organism can succeed or fail in doing something. Although I can successfully or unsuccessfully confide and thus can exercise that capacity correctly or not, I cannot burn to a crisp successfully or unsuccessfully. Capacities are thus potentialities whose exercises are by nature subject to a standard. In my terminology, a capacity is such that a principle describes its nature and is thereby normative for its exercise. An exercise is correct to the extent that and because it meets that principle by possessing the properties mentioned in the principle, incorrect to the extent that and because it does not.

2.2

Since practical and theoretical reason can be exercised correctly or incorrectly, they are capacities with principles that describe their nature and are thereby normative for their exercises. On this metaphysics, rules of rationality are whatever partially specifies the nature of those capacities. Here is why.

To follow the instrumental rule of modus ponens, say, is to exercise the aspect of the mind that distinguishes us from the other animals. It is to exercise reason. If an exercise in part consists in following such a rule, reason is in part a capacity to follow that rule. To be a rule of rationality, then, is partially to specify the nature of reason. So a rule of practical rationality partially specifies the nature of practical reason, a rule of theoretical rationality the nature of theoretical reason.

Practical and theoretical reason are thus capacities whose exercises are correct to the extent that and because I meet the rules of rationality by possessing the properties mentioned in those rules. I do not mean that to exercise them correctly is to follow a set of rules picked out by their content. The constitutivist metaphysics identifies rules of rationality by their role in the account of a self-conscious capacity. Their content depends on what practical and theoretical reason are potentialities to do.

3.1

If practical reason is our will, its principle is a specific determination of the one that characterizes a will in general just as the function of our heart is a specific determination of the generic function of a heart. Hearts circulate blood. Ours does so in a specific way. What, then, is a will? How does ours do in a specific way what a will in general does?

An exercise of a will is complete only when the animal acts. Yet an animal succeeds not by simply acting but by pulling off the action. Take my cat on his way toward me on the settee. He succeeds when he jumps up, kneads me, and lies on top of me. His exercise of his will is complete only when he finishes doing what he is doing such that he is no longer φ-ing and has φ-ed. He can fail. Maybe he misses the jump, or perhaps I toss him aside because he kneads too long or roughly. He is no longer φ-ing but has not φ-ed. He was φ-ing but did not φ. But maybe this time he succeeds, no longer acting but now having acted. His exercise of his will is only then complete. Acting as he wills to act is thus part of the standard for exercises of his will. Given the constitutivist metaphysics, it is part of the nature of his will. As with his will, so with any. As we might put it, a will is in part a capacity to realize the object of its representations.

As with the feline will, so with the human. Say I aim to read The Tunnel at the pace at which William Gass composed it. I succeed only if I finish the book twenty-eight years after starting it. My exercise of my will is complete only when I am no longer φ-ing and have φ-ed. I can fail. Maybe someone casts all copies to the flames. I am then no longer φ-ing but have not φ-ed. I was φ-ing but did not φ. But perhaps this time I succeed, no longer acting but now having acted.

Of course, I differ from my cat. I am self-conscious, which transforms the nature of the activity and capacity. This self-consciousness is not grand. My kid nephew is an exemplar. Any game involves house rules that he enforces while exploiting opportunities they make possible. That is self-consciousness in action. It is why I can follow a rule and know what I am doing and why. It is why ignorance is an error. A full account of our will must explain this self-consciousness, of which I provide only a part in the next sub-section. Even absent that explanation, though, if practical reason is our will, its exercise is complete only when I act as willed just like any other type of will. Acting as I will to act is thus part of the standard for exercises of practical reason. Given the constitutivist metaphysics, it is part of the nature of practical reason. As we might put it in order to mark its distinctive self-consciousness, practical reason is in part a capacity to realize concepts.

I model this account of the relationship between the generic will and practical reason on Kant's account of the relationship between the generic faculty of desire and the will. He claims that ‘the faculty of desire is a being's faculty to be by means of its representations the cause of the reality of the objects of these representations’ (Reference Kant and Gregor1997: 5:9n). He also claims that the ‘faculty of desire whose inner determining ground . . . lies within the subject's reason is called the will’ (Kant Reference Kant and Gregor1996: 6:213). As I say that animals have wills and ‘practical reason’ is the name for the self-conscious type, so he says that animals have faculties of desire and ‘the will’ is the name for the rational type. Terminology differs, but the view is the same. I modify his terminology only because the language of desire might connote passivity in a way that the language of a will, or of volition, does not.

3.2

An exercise of practical reason succeeds only if I get something done. Thought is enough for God to shape the world and for me to know it. For me to shape the world, though, requires more. Since this difference between practical and theoretical reason matters for the answers to the questions about the content of the instrumental rule, I explain it below.

An exercise of theoretical reason about a donkey chomping on a soup can, say, is correct only if such a beast of burden is chewing a can. Whether that is so does not depend on my thought about her. The object of my theoretical representation is thus independent of that representation. Because theoretical reason is a self-conscious capacity, I recognize this independence in a correct exercise of it. That is why it is a mark of madness to think that what that donkey is up to depends on my thought about her.

That mark of madness in an exercise of theoretical reason, though, is a mark of maturity in an exercise of practical reason. Such an exercise about confiding in you is correct only if I tell you a tale by which you come to know my mind. Whether I am confiding depends on my representation of what is happening. You might come to know my secrets as sounds escape in my sleep, but that no more constitutes confiding than cheating with counterfeit cash constitutes commerce. I am not acting because those sounds do not realize a representation. You might instead come to know my secrets as I recite them without knowing about the microphone broadcasting to the school. I am acting but not confiding because although those sounds realize a representation, it is the wrong one. In contrast, if I summon the trust and courage to tell you what I have so long been afraid to say, I am confiding. I realize that concept. The object of my practical representation thus depends on that representation. Because practical reason is a self-conscious capacity, I recognize this dependence in a correct exercise of it. I recognize that what I am up to, and indeed whether I am up to anything, depends on my thought about it.

The thinking that partially constitutes an exercise of practical reason is thus about the doing that is the other part of that exercise. The doing that partially constitutes an exercise of practical reason realizes the thinking that is the other part of that exercise. Each is what it is because of its relationship to the other. Part of what it is to be self-conscious is to understand this relationship in a way that grounds responsibility for securing it and failing to do so in an exercise of my will. For practical reason to be in part a capacity to realize concepts, then, is for thinking and doing to be interdependent parts of its exercise. It is for doing to realize thinking about doing—for action to realize thought about action—at least when I exercise that capacity well (for more on the transformative nature of self-consciousness, see Boyle Reference Boyle2016; Marcus Reference Marcusmanuscript).

4.1

I aim here to co-opt an explanation that some philosophers offer for the standard interpretation.

Kant claims that the hypothetical imperative says that ‘whoever wills the end also wills (insofar as reason has decisive influence on his actions) the indispensably necessary means to it that are within his power’ (Kant Reference Kant and Gregor1998: 4:417). That appears to be the standard interpretation. His explanation of the analyticity of this imperative, though, gives the lie to that appearance. He claims that this imperative merely ‘extracts the concept of actions necessary to this end merely from the concept of a volition of this end’ because ‘in the volition of an object as my effect, my causality as acting cause, that is the use of means, is already thought’ (Kant Reference Kant and Gregor1998: 4:417). In effect, to will an end is to set about realizing it, which grounds the hypothetical imperative. To realize an end, though, is to take sufficient means to it, which involves taking any necessary means but requires more. What is already thought in willing an end is thus taking sufficient means, not merely intending necessary ones. Given this explanation of the analyticity of this imperative, I doubt that Kant is using the contemporary notion of a necessary means. I suspect that he means what my boss means when he says ‘Do whatever is necessary to get the job done’, which order I carry out only by taking sufficient means. (Kant goes on to argue that the syntheticity of propositions about means to specific ends does not threaten the analyticity of the hypothetical imperative. Since propositions about sufficient means are as synthetic as ones about necessary means, this argument fits with an imperative about either type.)

Since this essay is not an exercise in Kant interpretation, though, I leave off on these exegetical issues. The important point for present purposes is that contemporary arguments suffer from a similar slip. Consider this argument from Stephen Finlay, which he presents as a scrubbed-up version of Kant's argument:

However, if ‘intend the end’ ‘partially describes’ an intention and if ‘to intend the end is to intend to act so as to bring the end about’, the only way to ‘complete’ that description is with ‘intend sufficient means’. After all, necessary means are not enough to bring about most ends. Finlay is thus wrong that ‘to intend the necessary means is to intend to act so as to bring the end about’ (Reference Finlay and Pigden2009: 163). When necessary means are insufficient, to intend them is to intend partially to bring about the end. To intend to act so as to bring (fully) about the end is instead to intend sufficient means. Since any set of sufficient means includes any necessary means, the description in terms of ‘intend the end’ and ‘intend sufficient means’ is complete without help from ‘intend necessary means’. Further, some ends lack necessary means. I can get your attention, say, by shouting, jumping, raising my hand, throwing my hat, or doing whatever else might catch your eye. There is nothing I must do.

Hence, while ‘intend necessary means’ does not contribute anything new to the combination of ‘intend sufficient means’ and ‘intend the end’ when an end has necessary means, it does not contribute anything at all when an end lacks such means. It is at best redundant and at worst irrelevant. That is, in effect, my argument in this section.

4.2

Is this rule about necessary or sufficient means?

If practical reason is our will, this rule is about sufficient means. When an end lacks necessary means, there are none to take. When necessary means are insufficient, taking them is not enough to act as willed. Such an exercise is incomplete. I can complete it only by taking sufficient means. The principle of practical reason thus needs a rule about sufficient means. Otherwise, I can comply with the rules of practical rationality without acting as willed.

That argument might seem to imply only that a rule about sufficient means augments one about necessary means in the principle of practical reason. The first, though, obviates the second. A set of sufficient means includes any necessary ones. Although taking necessary means is not enough to meet that correctness condition, taking sufficient means is in part because I thereby take any necessary means. Hence, once the principle of practical reason includes a rule about sufficient means, there is nothing for one about necessary means to explain.

Can we salvage the standard interpretation by claiming that when there are multiple sets of sufficient means to my end, it is necessary that I take one of them? No. The proposition it is necessary that I take some set of sufficient means is not the proposition there is some set of means that it is necessary for me to take. The latter proposition characterizes the notion of a necessary means in the standard interpretation. The former proposition has nothing to do with that notion of necessary means. In fact, it is just a long-winded way of stating a rule about sufficient means. ‘Necessary’ is here a deontic modal. The proposition just says to take sufficient means. Why? Because otherwise I will not pull off the end.

4.3

Is this rule about intending or taking means?

If practical reason is our will, this rule is about taking means. Intending them is not enough to act as willed. Such an exercise is incomplete. I can complete it only by taking means. The principle of practical reason thus needs a rule about taking means. Otherwise, I can comply with the rules of practical rationality without acting as willed.

That argument might seem to imply only that a rule about taking means augments one about intending them in the principle of practical reason. The first, though, obviates the second. Thinking and doing are interdependent parts of an exercise of practical reason. Thinking is about doing, which realizes thinking. Although thinking is not enough for me to meet that correctness condition, doing is in part because I can do only if I think. That is to say that I can take means only if I intend them because of the way that what I am doing, and indeed whether I am doing anything, depends on my thought about it. Hence, once the principle of practical reason includes a rule about taking means, there is nothing for one about intending means to explain.

4.4

Is this rule about believed or actual means?

If practical reason is our will, this rule is about actual means. Whenever believed means are not actual means, taking them does not help, and might well hinder, my acting as willed. Such an exercise is at best incomplete. I can complete it only by taking actual means. The principle of practical reason thus needs a rule about actual means. Otherwise, I can comply with the rules of practical rationality without acting as willed.

That argument might seem to imply only that a rule about actual means augments one about believed means in the principle of practical reason. The first, though, obviates the second. Thinking and doing are interdependent parts of an exercise of practical reason. I act in some way to achieve an end only if I so intend. I must believe that I can achieve that end by acting that way, though, to so intend. I cannot sing your favorite song at karaoke to woo you unless I believe that I can woo you by so serenading you. Although taking believed means is not enough for me to meet that correctness condition, taking actual means is in part because I take means only if I believe that they are means given how what I am doing, and indeed whether I am doing anything at all, depends on my thought about it. Hence, once the principle of practical reason includes a rule about actual means, there is nothing for one about believed means to explain.

4.5

To sum up, if practical reason is our will, my exercise of it is complete only if I act as I will to act. To act as I will to act is to take, not merely intend, actual, not merely believed, sufficient, not merely necessary, means to my end. Since explicitly assuming that practical reason is our will leads to that alternative interpretation, the standard one reflects an implicitly assumed separation of practical reason and the will.

Once the principle of practical reason includes a rule about taking actually sufficient means, one about intending believed necessary means is at best redundant. The former can explain anything the latter can, and the latter cannot explain everything the former can. Moreover, the latter excludes most of our practical activity from the ambit of reason. Michael Bratman is thus in a sense right and in a sense wrong to say that the standard interpretation is ‘a central aspect’ of the requirement ‘that an agent fill in her plans with one or another sufficient means’ (2018: 53n2). It is an aspect of that broader rule. There he is right. It is not ‘central’, though, because its application is limited while that of the broader rule is not and its explanation is wholly derivative of that of the broader rule. There he is wrong.

5.1. Against Sufficient Means

Sarah Paul thinks that willing sufficient but not necessary means cannot be part of an exercise of practical reason because it cannot ‘be . . . regulated with respect to some standard of correctness . . . regarding the relation between means and ends’ (Reference Paul2013: 295). That standard ‘regulates’ that relation when (1) acting in some way is correct to the extent that and because it contributes to achieving the end and (2) I will to act that way because of the end, where that is the ‘because’ of rational basing. She claims that following a rule about sufficient means cannot meet that second condition because whenever there is a set of sufficient but not necessary means, there is at least one other. They are ‘equally rational ways of’ acting with respect to the end, ‘reason is mute as to’ which to will, and willing one is thereby ‘plumping’ (Paul 2103: 295). She thus claims that since a rule about sufficient means does not ground willing sufficient but not necessary means M rather than sufficient but not necessary means N, I cannot will M because of E.

That argument, though, is unsound. I can will M because of E without grounds to will M rather than N. When Paul says that ‘reason is mute as to’ which set of means to will, she means that reason says the same thing about them. Each is better than any other way of acting according to the standard that relates them to E. Each is a correct set of means given this end. This equality implies that this standard does not say of either that I must will it. That is compatible, though, with it saying of each that I can will it. Just as I can follow a law that permits rather than requires, so I can follow a rule of rationality that permits rather than requires. Just as I comply with a law in giving you a £10 note in order to buy the book without grounds for paying with a tener rather than two fivers, so I follow a rule of rationality in willing M because of E without grounds for willing M rather than N.

One way to understand this argument is to compare practical inference with theoretical inference. If I infer Q from Q and R if P and P, whether modus ponens regulates this inference depends on whether I believe Q because Q and R if P and P. Whether I could infer R on the same basis is irrelevant to the correctness of that inference. Similarly, if I infer M from E and E by either M-ing or N-ing, whether the instrumental rule regulates this inference depends on whether I will M because E and E by either M-ing or N-ing. Whether I could infer N on that basis is irrelevant to the correctness of that inference. In both cases, whether I base my conclusion on my premises instead depends on only the relationship between the conclusion and the premises.

Equality thus does not threaten regulation and does not undermine an interpretation of the instrumental rule in terms of sufficient means, at least without other assumptions. Which assumptions? An implicit denial that practical reason is the will. Paul says that the problem with willing sufficient but not necessary means is that ‘it requires no further judgment of choiceworthiness’ (Reference Paul2013: 296). The ‘further’ is important. The claim is not that M is not choiceworthy given the end but that M is not more choiceworthy than N given the end. Paul infers that willing M is thereby not part of an exercise of practical reason. That inference is sound, though, only if you assume that an exercise of practical reason consists in, or at least tracks, only judgments about what there is most reason to do, where ‘most’ excludes ties. This assumption is at odds with the correctness conditions of the exercise of a will and thus implicitly separates practical reason from our will.

5.2. Against Taking Means

Kolodny claims that rules of rationality ‘demand that our attitudes be related to one another in certain ways’ and focuses on the ‘special class’ of ‘requirements of formal coherence as such’ (Reference Kolodny2008: 366). He never explains why they concern only attitudes. He says that ‘it seems relatively uncontroversial that rationality is a kind of coherence or unity. So it is relatively clear how we might settle questions about what rationality requires; it is whatever is necessary for coherence’ (Kolodny Reference Kolodny2005: 511). Even if true, though, why need these rules only concern attitudes? Perhaps acting as I intend is coherent, failing to do so incoherent. Why not?

Kolodny explains his focus on formal coherence in a way that might explain the restriction to attitudes. He divides ‘requirements governing relations among attitudes independently of anything beyond those attitudes’ into those that require formal coherence and those that require informal coherence (Kolodny Reference Kolodny2008: 392n2). Although all meet his stipulative definition of a rule of rationality, he claims that requirements of informal coherence ‘stretch our ordinary attributions of “irrationality”, which are restricted to cases in which the subject is more immediately at odds with himself’ (Reference Kolodny2008: 392n2). Perhaps something is a rule of rationality, then, only if violations can license ordinary attributions of irrationality. Although he never considers whether such a rule can concern anything other than attitudes, he might reject this possibility on those grounds.

That argument, though, is unsound. To the extent that there is a well-formed class of ordinary attributions of irrationality, failing to act can be irrational. ‘You know exactly what to do and yet you are not doing it. That's irrational!’ is about what you fail to do and yet is not for that a category mistake. Anyone who teaches or writes philosophy is familiar with this type of criticism. Few will say that they did not intend everything not done. I doubt that you will revoke the charge if I explain that although I am not writing or reading or reflecting or ruminating even though the time is right, I intend it all. This attribution of irrationality is as ordinary as any because I am as at odds with myself as anyone. Yet it concerns my failure to take means.

Kolodny might reply that despite appearances, this attribution concerns intending means rather than taking them because ‘an action comprises an intention-in-action and the corresponding bodily movement, and the relation between an intention-in-action and the bodily movement is not . . . a rational relation. It is purely causal’ (Reference Kolodny2005: 548n36). He might claim that attributions of irrationality fundamentally concern the intention. He can say that although the whole action can inherit that property because one part possesses it, the other part does not possess it, whether derivatively or fundamentally. That is to say that the thinking that partially constitutes the action can be fundamentally irrational, the action can be derivatively irrational, but the doing that also partially constitutes the action cannot be in any way irrational. It is not a failure within the house of reason.

This argument invokes the causal theory of action, which decomposes human action into an inner mental part within the reach of reason and an outer bodily part that is not. This invocation is illegitimate. For one thing, this decomposition is a revisionist philosophical theory, not a bit of ordinary language philosophy. Ordinary attributions of irrationality do not presuppose it. For another, this decomposition is incompatible with the view that practical reason is our will. After all, (pure) efficient causation relates distinct existences whose essences are constitutively independent of each other. The thinking and doing that jointly constitute an exercise of practical reason then cannot each be what it is because of its relationship to the other. If practical reason is our will, though, doing is as much a part of its exercise as thinking because the relationship between them is rational, not (purely) efficient causal. This is why philosophers developing Aristotelian and Kantian trends in practical philosophy often reject the causal theory (Boyle and Lavin Reference Boyle, Lavin and Tenenbaum2010; Lavin Reference Lavin2014; Ford Reference Ford2014; Valaris Reference Valaris2015). Yet then failing to take means can be fundamentally irrational and license an ordinary attribution of irrationality.

5.3. Against Actual Means

Maybe the conflict with ordinary attributions of irrationality is with a rule about actual means. If my evidence is misleading, I might not take actual means to my end. If my belief about the means is false through no fault of my own, though, I am not ‘at odds with myself’ but with the world. An attribution of irrationality is illegitimate. Hence, if something is a rule of rationality only if every violation licenses an attribution of irrationality, the alternative interpretation of the instrumental rule is false. (You can make the same objection against the other aspects of that interpretation. After all, when I do not take actually sufficient means through no fault of my own, I do not take actually sufficient means through no fault of my own. I ignore this complication because my response generalizes in an obvious way.)

No rule of rationality, though, is such that every violation licenses an attribution of irrationality. Take modus ponens. I might believe that P, believe that Q if P, and yet not infer Q. Whether I am irrational depends on why I do not infer. Maybe although I believe Q if P, I come to believe P when the conditional is out of mind. Say on Monday I make a date with someone for Friday and thus believe, indeed know, that if it is Friday, I have a date with him tonight. Dinner, drinks, dancing, and all that jazz. Say on Friday morning, I believe, indeed know, that it is Friday. I am then in a position to correctly infer and thereby know that I have a date with him tonight. Yet I might not. (Koziolek [Reference Koziolek2018: 2–4] discusses the importance of this case for an account of belief.) Perhaps it slips my mind because of the whirl of life. Students to tutor, lectures to give, meetings to attend, essays to mark, tasks to administer, and so on might so consume my time, attention, and energy that I never think beyond the next moment. Perhaps something more dramatic happens, like the loss of a limb or a loved one. I might then not infer that I have a date tonight without being irrational and while knowing the propositions that license this inference. After all, if, in the middle of it all, I think ‘Wait. I made a date with him for Friday, and today is Friday. I have a date tonight!’ and don my dancing shoes, I infer from grounds already possessed. If I do not, though, I violate modus ponens. If I do not show up to dine and drink and dance, I violate the instrumental rule. Yet ordinary attributions of irrationality are inappropriate if the explanation for why I violate it excuses. (For what it is worth, these types of excuses also show that some violations of the standard interpretation do not license ordinary attributions of irrationality.)

Ordinary attributions of irrationality do not track violations of rules of rationality. They track culpable violations. Every violation licenses an attribution of irrationality, then, only if every violation is culpable. As far as I can tell, every violation is culpable only if rules of rationality include awareness conditions such that they apply only when I am simultaneously occurently aware of every attitude and action (or lack thereof) that they concern (see Lee Reference Lee2018). Whether these awareness conditions make sense in another account of these rules, though, they are incompatible with the constitutivist metaphysics. Given this metaphysics, a rule of rationality constitutes a correctness condition for the exercise of a self-conscious capacity. If practical reason is the will, one of these conditions is to act as willed. I violate that rule if I do not take actually sufficient means to my end, which can happen even if I am never simultaneously occurently aware of every attitude or action (or lack thereof) that it concerns. Forgetting my date does not exempt me from that rule when I do not show up even though I can only take sufficient means to it by walking there then. Every violation of a rule of rationality licenses an ordinary attribution to irrationality, then, only if practical reason is not the will.

5.4

Practical philosophers cannot put aside their differences in order to discuss the instrumental rule because its formulation depends on views about the nature of our agency. Interpretations of it are as disputable as is anything else in practical philosophy. To endorse one is to take a stand on whether the reach of reason extends only to thinking or beyond to doing. The standard interpretation implies that our agency is fundamentally the potentiality to think, the alternative that it is fundamentally the potentiality to realize thought in action. Which is correct depends on our nature. Fundamentally, can we only think, or can we do as well? If practical reason is our will, our agency is fundamentally a potentiality to realize concepts or to act from thought about action. The instrumental rule codifies an aspect of that potentiality by saying to take actually sufficient means to my ends.

6.1

Is success in an exercise of practical reason bound up with following this rule? ‘No’ might seem like the answer. If my end is immoral or imprudent, taking means to it is incorrect. Whether taking means is correct thus depends in part on whether having the end is correct. Yet since the instrumental rule does not discriminate between ends, it might seem to imply that taking means to incorrect ends is correct. If so, it is extensionally inadequate and lacks normative status.

The basic structure of the objection is clear, its details and soundness less so. They depend on assumptions about the nature of normativity. For example, if you assume that facts about reasons exhaust the fundamental level of normative reality, the objection is that since the instrumental rule does not discriminate between ends supported by reasons and ends not, it does not track reasons to act. It implies that I have sufficient reason to take means when I actually have decisive reason not to act that way because I have decisive reason against the end. If you instead assume that facts about goodness exhaust the fundamental level of normative reality, the objection is that since this rule does not discriminate between good and bad ends, it does not track which ways to act are good. It implies that taking certain means is good when in fact acting that way is bad because the end is bad. Anyone who accepts some such account of the nature of normativity must deny the normative status of this rule, explain what the objection gets wrong, or accommodate the objection by weakening the rule. Kolodny (Reference Kolodny2005) denies the normative status of this rule. John Broome (Reference Broome2007), at least at one time rejected the objection by interpreting this rule as a wide-scope norm that requires that I either take means or give up my end. Kieran Setiya (Reference Setiya2007) rejects the objection by claiming that this rule is a rule of theoretical rationality with epistemic but not practical normativity. Mark Schroeder (Reference Schroeder2005) weakens the rule by claiming that it grounds a reason to take means that is defeated when I have incorrect ends. There are many other examples.

Those are the options, though, only if the assumed account of the nature of normativity legitimizes the objection. The constitutivist metaphysics does not. On it, the instrumental rule partially specifies the nature of practical reason. Exercises of that capacity are thereby subject to a standard that includes that rule. An exercise is thus correct only if I take sufficient means. That rule does not specify other properties an exercise needs in order to be correct, but it is compatible with further correctness conditions. If there are such further conditions, there are other rules that also partially characterize practical reason.

This objection assumes that there are correct ends that partially determine how to exercise practical reason. It thereby assumes that there is a rule of rationality that concerns ends that partially specifies the nature of that capacity. That rule is as much a part of the standard for exercises as is the instrumental rule. Meeting it is as necessary to the correctness of those exercises as is meeting the instrumental rule. To exercise practical reason in a way that violates it is thus incorrect according to the normative standard of which the instrumental rule is a part. The fact that the instrumental rule does not discriminate between ends is then irrelevant. There is no extensional inadequacy and no illicit bootstrapping.

The only other way to try to legitimize this objection within the constitutivist metaphysics is to deny that there are rules of practical rationality about ends. On that metaphysics, there are then no correct or incorrect ends. The fact that the instrumental rule does not discriminate between ends is again irrelevant, this time because there is no distinction to track. There is no extensional inadequacy and no illicit bootstrapping.

Assume the constitutivist metaphysics, then, and the objection is self-undermining. If there are incorrect ends, the aspect of practical reason that explains why also explains why an exercise in which I take sufficient means to them is incorrect. If there are no incorrect ends, an exercise in which I take sufficient means to any end is correct.

6.2

Since the explanation for why that objection is self-undermining invokes another rule of rationality, the significance of the instrumental rule might seem doubtful. Why not think that the rule that explains why taking means to incorrect ends is incorrect also explains why taking means to correct ends is correct? That other rule seems to do all the work. In what way, then, does the instrumental rule tell me what to do?

The answer turns on understanding a difference between the question about the status of this rule given the constitutivist metaphysics and the question about its status given other accounts of the nature of normativity. Although I shall not discuss other accounts in detail, they understand the question about its status as about the truth of the conditional ‘I exercise practical reason correctly if I follow the instrumental rule’. After all, only then do instances in which taking means is incorrect challenge that status. Assume the constitutivist metaphysics, though, and the question about its status is about the truth of the conditional ‘I exercise practical reason correctly only if I follow the instrumental rule’. An exercise that does not meet this rule is incorrect at least in part because I do not take sufficient means. An exercise in which I take such means, in contrast, is thereby to that extent, but only to that extent, correct. Likewise for other rules of practical rationality. Taking sufficient means is thus a perfection, failing to do so an imperfection, in the exercise of practical reason.

Perhaps an analogy helps. Consider a recipe. Each step tells me what to do. To make the dish correctly, I must meet each correctness condition. If I bungle the first step, that dish is ruined. The irrelevance of the third step in that instance, though, does not show that it is never significant. If I pull off the other steps but muck up it, this dish is ruined. Yet the other steps do not explain why. Similarly, if the recipe has five ingredients, buying the wrong type of the first is enough to ruin it regardless of whether I buy the right or the wrong type of the others. That, though, does not show that the third ingredient is insignificant. If I buy the right type of the others but the wrong type of it, the recipe is ruined. Correctness requires buying all the right ingredients and following all the steps in order.

An exercise of practical reason is likewise correct if and only if and because I meet the rules of practical rationality. An incorrect end undermines an exercise. The irrelevance of the instrumental rule in that exercise, though, does not show that it is never significant. Having a proper end but not taking sufficient means also undermines an exercise. Yet a rule concerning ends does not explain why. Having a good end is essential to exercising practical reason well but it is not enough to do so, as is taking sufficient means.

I am here aping a form of explanation that certain philosophers inspired by Aristotle and Kant use to explain the unity of the virtues. For example, John McDowell claims that ‘the particular virtues are not a batch of independent sensitivities. Rather, we use the concepts of the particular virtues to mark similarities and dissimilarities among the manifestations of a single sensitivity, which is what virtue, in general, is’ (Reference McDowell2001: 53). A specific virtue term, then, picks out an aspect, its corresponding vice term the lack of that aspect, of a virtuous character. Possessing each aspect is necessary but insufficient for a virtuous character, and lacking any is sufficient but not necessary for the lack of a virtuous character. Possessing them all is jointly sufficient for a virtuous character.

McDowell emphasizes the way in which ‘no one virtue can be fully possessed except by a possessor of all of them’ (Reference McDowell2001: 53). Edward Harcourt (Reference Harcourt2016: 227–33) points out that the possibility of possessing aspects of a virtuous character to an incomplete degree implies that correct action has ‘layers’ that can come apart in cases of incompetence or wickedness. Given that a virtuous agent is someone who acts correctly and given that the instrumental rule is part of the principle of our capacity to act, this rule is part of that story. It is a layer that can come apart from others in imperfect actions but that is part of every perfect action. More straightforwardly, you can meet it and yet act imperfectly, but you cannot act perfectly without meeting it because taking sufficient means is part, though only part, of what makes an action correct or virtuous.

Korsgaard connects this view of the unity of virtue with a view of the unity of the rules of practical rationality. She claims that there ‘is really only one virtue, but there are many different vices, different ways of falling from virtue, and when we assign someone a particular virtue, what we really mean is that she does not have the corresponding vice. In a similar way, there is only one principle of practical reason, the categorical imperative viewed as the law of autonomy, but there are different ways to fall away from autonomy, and the different principles of practical reason really instruct us not to fall away from our autonomy in these different ways’ (Korsgaard Reference Korsgaard2009: 71–72)

Each rule of practical rationality partially specifies the principle of practical reason just as each virtue partially specifies that single sensitivity. Jointly those rules completely specify that principle just as jointly those virtues completely specify that sensitivity. With respect to each rule, meeting it is necessary but insufficient for the correctness of the exercise just as with respect to each virtue, possessing it is necessary but insufficient for the virtuousness of my character. With respect to each rule, failing to meet it is sufficient but not necessary for the incorrectness of the exercise just as with respect to each virtue, failing to possess it is sufficient but not necessary for the lack of a virtuous character. Meeting every rule is sufficient for correctness because jointly they fully specify the principle of practical reason just as possessing every virtue is sufficient for a virtuous character because jointly they fully specify that sensitivity.

There is thus only one principle of practical reason, which consists of all of the rules of practical rationality. It is, and thus they are, thereby normative for the exercise of practical reason. Although not every exercise that meets the instrumental rule is correct, every correct exercise is so in part because the agent takes sufficient means. Likewise for any other rule of practical rationality. The instrumental rule thus has the same normative and explanatory status as any rule of rationality, at least on the assumption of the constitutivist metaphysics of practical reason understood as our will.