Humeans interested in the connection between psychology and action are distinguished by the view that having desires, which are separate from beliefs, is necessary for an agent to have a motive to action. Their view is traceable to an argument in Hume’s A Treatise of Human Nature (1739–1740) where he maintains that reason alone does not motivate, and that passions (desires) are required to prompt action. For Humeans, accordingly, motives are constituted by desires in conjunction with beliefs, and only this combination of mental states can explain action.

Some Humeans, although not Hume himself, attempt to offer a theory of reasons for action. They are convinced that an adequate action theory should capture what it is for a person to act from reasons. Typically, reasons for action are thought to be of two kinds: motivating reasons and justifying reasons. Motivating (or explanatory) reasons explain why agents are motivated in acting, while justifying reasons offer considerations to show why actions are advisable in a moral or prudential sense.Footnote 1 Humeans offering theories of reasons for action are often criticized for an inability to account for justifying reasons. Since their method is observational and descriptive, they are committed to offering explanations of action in terms of desires that people actually possess. As Jonathan Dancy formulates this line of criticism, Humeans cannot explain how agents would ever be able to act on justifying reasons, or why they would ever be able to do anything on the grounds that it is right or recommended. Dancy and others think this is so because they see justifying reasons as objective states of affairs in the world, rather than as any of the subjective, psychological states that Humeans emphasize (Dancy 2000).

In the discussion that follows, my aim is twofold: first, to suggest ways in which Humeans can offer a psychological account of justifying reasons; and second, to argue that the anti-Humean line, which I shall call the “objective-reasons theory,” has its own challenge: it cannot explain how any particular action results from reasons as these philosophers characterize reasons. The objective-reasons theory has been defended by philosophers besides Dancy (2000, 2018), including Schueler (1995), Quinn (1993), Heuer (2004), Korsgaard (2009), Searle (2001), and Lord (2018). Some other philosophers, such as Alvarez (2010), Hieronymi (2011), Mantel (2018), and also Lord, have recently attempted to strike a compromise between Humeanism and objective-reasons theory, and I address their positions as well.

In Sect. 1 of this paper, I explain what I call the “leading problem” for Humeanism in action theory: the claim that normative practical reasoning proceeds with agents' considering the way the world is, rather than with considering their own desires. I also note some other pressing objections to the Humean view of reasons. In Sect. 2, I sketch a theory of Humean practical reason in an effort to reply to the leading problem. Section 3 then shows how objective-reasons theory has its own quandary in its inability to explain actions. In Sect. 4, I discuss the more recent attempts to deal with the leading problem, ones that have more plausibility than the traditional approaches. Section 5 then replies to the objections I noted in Sect. 1. In sum: The critics of Humeanism say that Humeanism cannot account for justifying or normative reasons; I show that it can. Then I argue that objective-reasons theorists cannot account for explanatory (or motivating) reasons.

1 The leading problem for Humeanism about action

To act on reasons, according to many of the objective-reasons theorists, is to respond to the way the world is. I eat a sandwich because I am hungry. I drive my car to the office because I have an appointment at 9 a.m., and I cannot arrive by then if I walk. As an agent in the world, I am not thinking about my own psychological states when I act, about my desires and beliefs, but rather about the circumstances around me. In his most recent book, Practical Shape (2018), Dancy writes:

… I don’t see any need for all practical reasoning to start from a desire that is already somehow given, the only question being whether or how one is to implement it. But practical reasoning does not occur in the void either. It is enquiry, but not idle enquiry; it is enquiry that serves a practical purpose. To have a practical purpose is to want something, or to have an aim. But when I reason in the service of an aim, I do not need to reason from having that aim. I reason to ways of achieving what I want, of realizing some aim that I have, but my reasoning need not start autobiographically, from the fact that I want this or have that aim (p. 123).

For Dancy and others, reasoning about the ways of fulfilling goals does not mean that agents think about any of their desires, but rather about the circumstances that favor their acting in certain ways. Practical reasons depend on objective features in the world. For instance, my duty as a professor to evaluate student papers fairly is not dependent on my desires. It depends on facts about the goals of teaching, the aims of evaluation, what my job is, and so on. So, the argument goes, if my desires and beliefs are the only reasons I can have for evaluating papers fairly, as the Humean view says, then I can never be motivated by the reasons that make that action right (Dancy 2000; Schueler 1995; Quinn 1993; Heuer 2004). My desires and beliefs are not the proper sort of thing to count as reasons, since they are psychological states.

I call this problem about normative reasons “the leading problem” for Humeanism. However, there are additional challenges posed by critics, and others might argue that these challenges are equally pressing. Among them are the following.

  1. (1)

    We have no reason to act on a desire to achieve an end if we have no reason that justifies the end (Korsgaard 1997, 1998; Raz 2011; Quinn 1993; Dancy 2000, 2018), i.e., desires do not justify ends (Schueler 1995).

  2. (2)

    When one acts on a desire, it is not the desire, but one’s belief about the desire, that does the work in explaining action. (Schueler 1995).

  3. (3)

    Doing something in response to a state of affairs counts as doing it for a reason. This requires no reference to beliefs and/or desires (Bittner 2001; Dancy 2018; Lord 2018). Reference to desire and belief is only necessary if one is already committed to such a theory; thus, Humeanism begs the question (Bittner 2001; Dancy 2018).

  4. (4)

    Humeanism cannot account for the fact that we decide which reasons to act on (Searle 2001; Korsgaard 2009).

I later address these challenges.

2 A defense of a Humean theory of reasons

The Humean Theory of Motivation says that a desire separate from belief is necessary to motivate action. A Humean theory of reasons for action is harder to typify, but it is safe to say that a theory of practical reason is Humean if it holds that the justification of action requires reference to desires, which play a part in causing action. Donald Davidson first promoted the idea, adopted by Humeans, that actions, which are intentional, can also be viewed as events causally explained by reasons composed of beliefs and desires (1963, pp. 685–690). In contrast, if reasons are facts “in the world,” then even if a theory of practical reason requires reference to facts about desires, the criticism that a justification could never motivate or explain seems correct. This is simply because desires and facts about desires are two sorts of things. Another way to understand the critique is this. When I reason about what I ought to do, I think about objects and states of affairs. Objects and states of affairs are the bases for the content of my beliefs and desires, but I do not think about my beliefs and desires themselves when I am reasoning. When I consider how I ought to dress, I take account of the fact that it is snowing, but I don’t think about my belief that it is snowing and about my desire to keep my feet dry. For objective-reasons theorists, we are moved by the objects of our reasoning rather than by our mental states.Footnote 2 Thus, Davidson and Humeans are wrong.

David Hume, the progenitor to the Humean theory of motivation, recognized a similar issue when he wrote about how we can be motivated by duty, the acknowledgement that an action is obligatory.Footnote 3 This seemed to pose a problem for his sentimentalism and for his virtue theory, since the thought that doing an action is the right thing to do is neither a sentiment nor a motive. Hume’s proposal, however, was that an unpleasant feeling provoked by noticing that one lacks natural motives like kindness and gratitude can move that person to do the right thing when those natural motives are absent. Thus, to be motivated by “a sense of duty,” according to Hume is to be motivated by a certain reaction to one’s motives, or lack of motives (Hume, Treatise 3.2.1.18).Footnote 4 Hume’s opponents may object to this proposal on the grounds that reference to an unpleasant feeling does not capture what a moral justification of action is. But on Hume’s account, this is what it is to think about duty, in the sense that one is comparing one’s actual motivational set to the set of motivations it is one’s duty to have. My point here is that on Hume’s view, I am paying regard to duty when I turn to my own desires and motivations and react to them. So, contrary to the point from Dancy and others, I am considering my own motivations rather than circumstances in the world.

In what follows in this section, I want to suggest that Humeans can offer an analogous account of practical reasons. First, I offer what I take to be a plausible rendition of the source of Humean practical reasons. Second, I briefly discuss two prominent Humean positions that I admire, but do not adopt. Third, I address what sort of thing a Humean reason for action is—a desire, another mental state, or an extra-mental state—in reply to the leading criticism.

2.1 The source of Humean practical reasons

My somewhat contentious view is that, on the Humean view, all desires for which there are identifiable means give us not just explanatory, but also justifying reasons for action.Footnote 5 It is also the case, however, that Humeanism is not committed to saying that we ought to do everything we have a justifying reason to do.Footnote 6 A single justification does not itself obligate. We experience multiple conflicting desires that give us justifications to act for various ends at any given time; however, we cannot act on them all at the same time. To determine on which among the conflicting desires we should act, we must sort out our own priorities or values. I contend that the process by which we ascertain which of our desires are of greater or lesser value to us is the origin of normativity for certain desires—namely for those that we find ourselves valuing more strongly over others. I explain further below.

I have argued elsewhere that having a certain attitude toward our own desires for ends constitutes valuing or caring for certain ends more than others. Moreover, this attitude establishes a qualitative assessment of our desires relative to one another (Radcliffe 2012). I here summarize my proposal for characterizing the source of normativity, on a Humean view.Footnote 7 Following philosophers like Frankfurt (1971), I have defended the idea that reasons should be designated by second-order approvals or disapprovals of an agent’s current motivating states. Even though virtually all of an agent’s desires are, on my view, reason-giving, an agent’s first-order desires often do not reflect what she or he cares more or most deeply about.Footnote 8 Frankfurt puts the point in terms of desires, but second-order approvals are, in motivational terms, for a Humean, equivalent to an agent’s desires to desire, and desires not to desire. It is crucial to distinguish in this discussion the quantitative intensity or strength of a desire from its qualitative value. When my alarm goes off, I may want more strongly to stay in bed than I want to jump up and get dressed, but I may value getting up at the same time everyday more than I value sleeping later. Or I may care about being on time for work (to which rising when the alarm rings is a means) more than I care about the comfort of staying in bed. Still, I stay in bed because my desire for that comfort is currently causally stronger, as I feel the warmth of the blankets against my body.

The second-order desire view has been the subject of criticism, with the chief issue being why second-order desires have some special reason-giving authority that first-order desires lack. This I address shortly. Another point frequently raised about a second-order desire view is that it leads to an infinite regress of desires of increasingly higher orders. My reply, however, is that our inquiry is done within a naturalistic framework based on normal human psychology. People do not typically develop desires about their desires to desire, much less anything higher level than that. It is simply difficult to get a coherent grasp of a third-order desire; having such a desire certainly does not come naturally and spontaneously to mind. So, I think that the infinite regress issue can in this context be discarded on the grounds that it is not descriptive of human psychology. Yet, another question is why we would always expect persons with harmful dispositions to disapprove of them. Warren Quinn poses the infamous example of a person with the quirky disposition to turn on all radios he encounters and notes that it is conceivable that this person simply would lack a second-order desire to rid himself of the disposition (Quinn 1993, p. 239). To use a more mundane example, it is conceivable that a chronic smoker might not disapprove of her disposition to smoke. So, these individuals have no reason to do anything differently on the second-order desire view. One response to the critique, which I have rejected, is to say that not all second-order desires are authoritative; the relevant second-order desires are the ones that designate the retention or acquisition of desires that are mutually consistent and that specify a coherent network of ends. Coherence, however, cannot answer the question why second-order desires in particular designate better reasons for action than first-order desires, since the first-order desires also may exhibit these features of coherence.

Now to the question about reason-giving authority. My view about the normative status of second-order desires borrows from Hume’s account in the Treatise and his reference to a general point of view (some call it the “moral point of view”). Hume is sometimes accused of lacking an account of normativity, but in fact, his theory offers an account of several sources of norms (see Beauchamp 2008). The Humean can hold that deliberation about the value of one’s desires must take a certain form to constitute proper deliberation. Just as the Humean (like Hume) can formulate norms for justifying a belief by looking at the natural process of belief formation, so she can formulate norms for evaluating desires by examining the natural process of formulating second-order judgments. While Hume takes experience as authoritative in his account of causal belief formation, he does not identify all patterns of experience as justifying the resulting belief. Rather, he formulates norms that derive from common practices and typical mental functions that allow us to acquire beliefs that are generally useful in ordinary life.

Among principles that govern the causal reasoning that underpins our beliefs in matters of fact are the following. (1) The cause and effect must be contiguous in space and time. (2) The cause must be prior to the effect. (3) There must be a constant conjunction between the cause and effect. (4) The same cause always produces the same effect, and the same effect never arises but from the same cause. (5) Where several different objects produce the same effect, it must be by means of some quality common among them. (6) The difference in the effects of two resembling objects must proceed from the features in which they differ. (7) When any object increases or diminishes with the increase or diminution of its cause, it should be regarded as a compounded effect, derived from the union of the several different effects that arise from the several different parts of the cause. (8) An object that exists for any time “in its full perfection” without any effect cannot be the sole cause of that effect; it requires assistance by some other principle that helps advance its operation (Treatise 1.3.15). So, we would be mistaken, for instance, to identify diet as the cause of heart disease, since to do so is a violation either of the seventh principle (with heart disease being the result of some combination of diet, heredity, exercise, smoking status, etc.), or of the eighth rule (since one might eat a fatty diet and not have heart disease).Footnote 9

Justified beliefs, which are judgments of facts, have the formal features of endurance, stability, and coherence, but someone who is consistently delusional or self-deceived may very well have a set of beliefs that exhibit these features. Such a network of beliefs is, however, idiosyncratic; it is not shareable. The general rules that govern belief formation are derived from a public or common perspective on experience. We find ourselves in a community of persons whose interactions depend on seeing the world in a common way. What we consider justified judgments about the world must be widely shareable in a such a way that anyone in the same situation with typical dispositions and the same type of experiences (and so who has in common a good number of other beliefs with persons around her) would come to adopt them as well.Footnote 10 The norms governing belief formation derive from the dispositions and means by which persons arrive at common beliefs. I think we can say that the regulations of good empirical judgment constitute a normative point of view on experience.

Analogously, one can articulate a Hume-inspired account of the normativity of practical judgments. Hume argues that moral distinctions and judgments are centrally dependent on sentiments or feelings, our psychological reactions to actions and the character traits they reflect. Hume bases his account of moral judgment on his observation that humans generally possess the mechanism of sympathy, the ability to feel some semblance of what people around them are feeling. We experience a natural sympathy with people affected, beneficially or harmfully, by the actions of agents in the world. We take agents’ actions as signs of their characters and approval or disapprove of them, depending on whether our sympathetic reaction to their effect on others is positive or negative (Treatise 3.1.2). As individuals, our natural sympathies are also affected by our proximity to people in space and time, and by our personal connections to them. We might feel more positively or more intensely about the accomplishments of friends and loved ones than we feel about similar actions on the part of people with whom we are little acquainted. Characters in history, whom we can only imagine, evoke our reactions, but often not as dramatically as those now living close to us. Yet, our judgments of the quality of the characters of the actors, when they have achieved similar results, are the same.

Hume explains these moral judgments, which may deviate from our initial natural feelings, as the result of our taking up a certain perspective, a general or common point of view, to “correct” for the variations in sympathies caused by our resemblance, contiguity, or causal connections to others. If we were to judge agents morally by our personal sympathies, we would have to deal with instability and conflict (Treatise 3.3.1.14-23). In judging the value of character traits, we judge the traits and the effects of the actions they produce, not according to our particular interests and our particular situation relative to the agent under consideration, but from a point of view others can occupy as well. Our judgments of morality, on Hume’s view, are derived from a common point of view, which he describes in terms of our sympathizing with the feelings of the people closest to the agent (under evaluation), rather than by appealing to our idiosyncratic feelings. If we sympathize with those directly affected by an ax murderer on the foggy streets of London in the 1800’s, we imagine the pain of the victims and respond as though we were present at that very scene, sympathizing with their reactions. Normally, people are able to do this to some degree and arrive at generally the same assessments of the agent as cruel, vicious, or heartless. Among the traits of others we judge in this way are all sorts of virtues and vices, such as gratitude and ingratitude, benevolence and malice, but also the virtue of acting for one’s long-term happiness, or prudence, and its opposite.

Hume’s account of moral judgment can, I think, be extended to offer a Humean story of personal deliberation about values and to derive standards that apply to such deliberation. To consider our first-order desires and their values to us, we psychologically withdraw from the way in which these desires feel to us at any given moment. We occupy a reflective point of view and regard them in light of such matters as their effects on our lives in the long run and their consequences for people around us, or for people we care about. This account of normativity is a hypothetical deliberative model. This is because conations with normative status for a person have their normative status regardless of whether the agent has undertaken the deliberation necessary to understand the normative force of certain desires. The model regards as having normative import not only those desires an agent does desire to retain after the deliberative process, but also those desires the agent would desire to retain if she engaged in the proper deliberation.

Consider the case of Marcel, a graduate student with an ailing mother. He fervently wants to concentrate on his graduate studies, but he also desires very much to be a support to his mom, who often needs to call on him for various kinds of help, including psychological, physical, and financial. How does he sort out his priorities and decide on what he has been reason to put his time? Such deliberation consists at least in part in imagining the long-range consequences of the alternatives for himself and for others intricately connected to him. First, he considers concentrating on his graduate studies. This would involve thinking both about the benefits of graduate life—the intrinsic enjoyment, his future in a profession, his sense of accomplishment—and the drawbacks—the hard work of study, lack of income, and shortage of time and energy and of other resources he might devote to his mother. On the other hand, he would consider how his life is affected by giving up his graduate school dream to tend to his mother’s needs, which would have both happy and unhappy consequences. He reacts to the resultant circumstances of the two options with approval or dissatisfaction, depending on his own dispositions, of course, but also from a point of view that disregards the immediate intensity or strength of his current first-order desires. In deliberation, Marcel sees his desires in a way that others could conceivably view them if they had adequate information about his life. Of course, they may not share his dispositions and interests (say, to be a loving son or to be a diligent student of history or to be steady and responsible person or to pursue novelty, or whatever else his dispositions might be). So not everyone would respond in exactly the same way. However, my claim is that to take up a deliberative perspective is to engage in a sort of weighing of options that has a certain structure, from which the normativity of the second-order desires is derived. It is controversial to depict Hume himself moving from the third-person spectator view in his moral theory to the first-person reflective perspective that allows us to judge our own characters. But it makes sense for Humeans to see reflection on our desires as taking a spectator’s view of our own motivations.Footnote 11

Consequently, there is a common, inter-subjective perspective from which we regularly make judgments of personal qualities, dispositions, and desires, both of our own and those of others. I call this a “practically normative” perspective, since it includes moral and prudential concerns (but is not restricted to those only). Hume himself notes that among the traits we approve in a general point of view are those motivations that conduce to an agent’s long-term interest and well-being, what we classify as traits of prudence. On this sort of naturalistic metaethics, moral and prudential or personal “oughts” are not of a different type, and they hail from the same source, even though we often speak of them differently. We can identify kindness and concern for another’s welfare as part of the moral arena, and an agent’s personal regard for her health as prudential, knowing that their values are rooted in the same sort of sentiments experienced from a shared, rather than from an idiosyncratic, perspective. It is also important to Humeanism that taking up the practically normative perspective is something we do naturally when we find ourselves with conflicting desires; it is not a process forced upon us by demands of morality or transcendent reason. It is rooted in our own nature.

Now I take up two questions about this account. One is whether it can guarantee that certain moral and prudential concerns will rank among those of higher importance than other concerns. For instance, would the fellow with the purported desire to turn on all radios he encounters find himself disapproving of that disposition when he reflects upon that goal from the normative viewpoint I have described? As I have argued elsewhere, I think the answer is: yes, if he is not in other ways very unusually disposed, not brain-washed or brain-impaired. After all, presumably, this desire thwarts many other desires he has, and it would also contribute to his long-term misfortune and that of others close to him. These are the sort of effects that are comprehended and considered by normal people from what I have identified as the reflective, practically normative point of view. Given the structure of reflection I have described and a person of typical or normal dispositions, many of what we ordinarily identify as important moral and prudential concerns will surely rank somewhere among a reflective agent’s most cherished ends.

The second question is what effect deliberation can have on an agent’s motivational set and actions. It is not plausible to think that the deliberative process somehow forces some of the agent’s first-order desires out of her motivational repertoire, or new first-order desires into it. We do not always end up doing what we think is best. The reflective process generates new desires of a second order—desires to desire new ends or desires to purge existing desires. If I am the sort of person within whom second-order desires have sufficient causal strength, then I might begin a program to work on my first-order desires. And it does makes sense to say that sometimes the strength of first-order desires can be affected by deliberation. However, in general, to say that a person desires (or would desire) to desire an end E after deliberation is not to say that she has a reason to pursue E now, without the relevant desire for E. It is to say that she has reason to acquire a desire for E. Often the steps required to acquire a desire are vastly different from the steps an agent would take were she to have the desire. If Jada enjoyed being with her parents and wanted to see them, she would visit them, if she has no conflicting commitments. But if she doesn’t enjoy the time with them and lacks the desire to see them, but does desires to desire visiting her parents, she might seek counseling to find out what the problem is and how she can change her feelings about interacting with them. Likewise, an agent might develop, through deliberation, a second-order desire to lose a certain first-order desire. She won’t necessarily lose it because of the second-order desire, but at the same time, if that second-order desire has enough causal strength, refraining from acting on the first-order desire is surely in part the means to expunge that first-order desire.Footnote 12

In summary, on the Humean view of reasons I here offer, if we were to engage in deliberation from the reflective perspective, and if no unjustified beliefs were to enter into the process, then the resultant attitudes are reason-giving evaluations. While, of course, there are differences in what individuals care most about, I contend that deliberation done according to the reflective process will yield a great deal of commonality among cherished ends because our shared human nature in part determines what we care about most.Footnote 13 Even the drug addict who has destroyed his mental and physical well-being and his personal relationships can plausibly be seen to care about some of these objects more than he cares about a drug high. But we cannot tell from his behavior, since he has been weak-willed in the past and now finds himself unable to quit.

2.2 Humeanisms

The literature offers several major attempts to defend a theory of Human practical reasons, among them accounts from Arpaly and Schroeder (2014), Blackburn (1998), Hubin (1999, 2003), Schroeder (2007), Sinhababu (2017), and Smith (1994). I agree with certain elements in each of them, obviously, although my view is not identical to any of them. My approach has most in common with Sinhababu’s.Footnote 14 I explain briefly why I put aside two of them.

The theory of reasons that results from Michael Smith’s famous Humean project is not, I think, thoroughly naturalistic, even though he argues otherwise (p. 186). On Smith’s theory, rightness consists in what fully rational beings would desire (p. 185). He maintains that his account is naturalistic in two senses: that the resulting right actions in the natural world would have no non-naturalistic features and that the realization of a fully rational being would require only a certain psychology, which is the subject of a natural science. When Smith asserts that the resulting right actions have no non-natural qualities, he means, I take it, that the property that a fully rational being would want actions to have is natural–for instance, the property of causing no pain or promoting happiness, or perhaps the property of being honest or being kind. The principle that explains why these properties are constitutive of right actions is that a fully rational being would desire them, and this principle itself is not the subject matter of natural science (as Smith notes). But the problem in claiming that the properties of right actions on his account are evidence of its naturalistic character is that, on pretty much any ethical theory, but perhaps Moorean intuitionism, the properties of right actions in the sense of which Smith is speaking are natural. After all, telling the truth, preserving one’s life, giving to charity, and developing one’s talents are right acts on a Kantian view; but none, as far as I can tell, has non-naturalistic features. They are all actions in the natural world. The principle that picks them out as right, indicating that they can be universally willed, is not itself a naturalistic principle, however. If one wants to argue that the rightness of these actions consists in the property of being universally willable, and that that is a non-natural property, then one must make a parallel claim about Smith’s analysis. One must say of Smith’s account that actions’ having the property of being desired by a fully rational being is a non-natural property as well. For what makes an act right on his account is not that it is desired by natural beings, but that it would be desired under ideal conditions. This feature, being desired under ideal conditions, is discernable only by reason—and, we should ask, by reason under what description? Certainly, not reason under a naturalistic description.

Nomy Arpaly and Timothy Schroeder’s argument for the Humean idea that practical reasons are not based in cognition, reasoning, or rational deliberation turns on their point that if acting for a reason always involves deliberation, this would lead to a regress, since deliberation is itself an action. Instead, they maintain that practical reasons depend upon “intrinsic desires”—desires whose objects the agent desires for themselves, not as a mean to any other end and not as a specific instantiation of an intrinsically desired end (2014, pp. 53–67). Their view, then, eliminates appeal to desires conditioned by ideal circumstances or ideally rational agents, second-order desires or approvals, and so on. While I am in sympathy with many of their points, I think that second-order desire accounts of how desires are related to reasons are less problematic than Arpaly’s and Schroeder’s, since people can have intrinsic desires for objects that they would acknowledge they do not really value or do not value as much as other objects they might have pursued instead. It is implausible to say that these people had reason, or best reason, to fulfill such desires. An alcoholic might pursue hard liquor as an end, but take no satisfaction in having it, and wish he or she did not care about it. Arplay and Schroeder do offer an analysis of such difficult cases and try to accommodate them within their theory, but I do not find the results very convincing, and it would take this discussion far afield to discuss this further.Footnote 15

2.3 The nature of a Humean practical reason and reply to the leading criticism

If Humean normative reasons are established by second-order desires in the way I have suggested, what sort of entities are they? Can they motivate, given the Humean theory of motivation, which insists that desire-belief pairs are necessary to motivation? Or is Dancy correct in his charge that, on the Humean view, an agent could never be motivated by normative reasons, since normative reasoning does not start with consideration of our desires, but rather with consideration of circumstances?

Since I have argued that normative reasons derive from second order desires, then such reasoning does at least sometimes take desires as its objects. Granted, in many cases, I am simply moved by desires that I am not thinking about. However, in tough cases, my reflection on what I really want is paramount to justification of one action among the alternatives, and sometimes these second-order desires are motivating. In such deliberation, I consider my first-order desires and whether I want to retain them or rid myself of them. However, it is not my considering my first-order desires that makes them reasons for me, on the Humean view. All of my first-order desires serve as reasons to take a means to the ends they give me, but some are better reasons than others; and this is what I discover when I find myself approving of some of them and wanting to retain or strengthen them. As I have noted, my desire to retain a desire might very well have some indirect effect on the motivational strength of the first-order desire. If I desire to desire more nutritious foods than I have been desiring (and eating), then my second order desire, if strongest among competing desires, may move me, say, to eat some broccoli, believing both that broccoli is more nutritious than the potatoes I’ve been eating and that eating a certain food can help me develop a taste for it. So, I might both desire to eat broccoli (as a means to acquiring a taste) and not desire to eat broccoli (because I don’t like it right now), but I have better reasons to eat it than not. The relevant second-order desire may have the motive force to affect my action, and I may eventually find broccoli appealing and desire it for itself. Then the second-order desire and the conflict go away. In another case, if a person most strongly wants to get rid of her desire to smoke, she may be motivated to try hypnotism or to start a program to assist tobacco withdrawal. Her desire to smoke may remain for a long time, and so she has a reason to smoke, but the Humean can say that she has better reason to undergo hypnotism or to undertake the withdrawal program than she has reason to smoke.

In the objective-reasons account, reasons are the considerations that persons think about when deciding what they ought to do. On the picture of Humean reasons I have offered, the objects of consideration are often desires, but it is not because they are objects of consideration that they are reasons. All desires that we can possibly act on are reasons, and our deliberations reveal to us which are more important. The Humean of this stripe starts with the theses that (1) desires are necessary to provide a motive impulse and (2) reasons for action require desire-belief pairs. She adds to these the notions that (3) second-order desires give us grounds for considering the objects of some desires as of more value than others, and (4) sometimes the second-order desires move us to undertake actions that the more valuable first-order desire would produce, if it were only strong enough. This is the answer to the question how normative reasons can motivate for the Humean. While all desires can give me (normative) reasons, I often think only about the circumstances when I act, rather than about my desires, especially when there are no perceivable conflicts of goals. While it is true that we never act without reasons, we sometimes act on the less-than-best reason, when the aim of the causally strongest desire is contrary to the aims of second-order desires.

3 The leading problem for the objective-reasons theorist

I show now that the anti-Humean line articulated by several philosophers has a difficulty concerning the explanation of action. The theory looks as though it can explain acting on reasons in terms of the circumstances for realizing our plans and goals. But the problem is that such a theory cannot explain why we do one particular action rather than another. Assume for the moment that practical reasons consist in states of affairs. Given the circumstances happening around me, I have reasons to do any number of actions at a given time. I cannot act on all the reasons presented by these circumstances, and my acting on some of them now may preclude acting on others in the future. For example, the fact that the submission deadline for this paper is approaching gives me a reason to continue working on it. But I also have reason to make a cup of coffee, since I have not yet had my second cup; and I have reason to entertain my cat, reason to check my e-mail, and reason to turn on the television, among other things. What would explain my acting on one of these reasons rather than another?

Some of the traditional objective-reasons theorists will reply that the answer is really quite simple: as an agent, I decide which reason to act on (e.g., Searle 2001; Korsgaard 2009, pp. 8–13). Searle says that acting on reasons requires that we suppose a gap between actions and their antecedents: a gap that is free will (p. 13). One way to recognize this gap, according to Searle, is to notice that you may have several reasons on which you could have acted, yet you act on one, and you know which one, without any investigation. “The reasons did not operate on you. Rather, you chose one reason and acted on that one” (p. 16). In answer to the question what fills the gap between the reason and the action itself, Searle replies: “Nothing fills the gap. You make up your mind to do something, or you just haul off and do what you are going to do, or you carry out the decisions you previously made, or you keep going, or fail to keep going, in some project that you have undertaken” (p. 17).

On Korsgaard’s view, the non-moral reasons we act on start with desires, which can be transformed into reasons. Her view is that “the reflective mind must endorse the desire before it can act on it, it must say to itself that the desire is a reason” (1998, p. 94). In doing so, the rational agent somehow creates out of the psychological state of desire something that is objective, that everyone can acknowledge as providing a reason for action. But reasons for action are normative; so Korsgaard needs to explain how we can fail to act on desires we endorse. Having reasons for action takes us only so far. We still need to follow through, and act on the reasons we have. But this issue is not my present focus. Presumably, a “fully rational” agent will reliably act on her desire-based reasons, at least when no weightier, moral reasons are overriding. So Korsgaard’s objective-reasons theory can easily get rational agents from having justifying reasons for action to acting on them in the ideal case. My concern instead is this. Accounts like the ones we are considering leave a mystery concerning the basis on which an agent would choose to act one way rather than another, as in Searle’s case, or would endorse one desire rather than another, as in Korsgaard’s. On the one hand, it might be that reasons are supposed to explain this, too. But then an infinite regress of reasons looms. Not only would we have justifying reasons for our actions, on the first order, we would also have second-order reasons for choosing or endorsing some of our reasons over others. Perhaps we would then also have still higher reasons, for sorting the reasons on the second order. Earlier, I argued that the charge of a regress of desires for the naturalistic view is answered by appeal to the fact that we do not naturally develop desires about our desires to desire. But on the objective-reasons view, the discussion is about on what basis we choose one set of considerations rather than another, and the search is not just for explanation, but for a justifying explanation. Justification (in a Kantian framework, as Korsgaard’s is) has formal requirements that are not answered by referring to what people do naturally. The implication is that the demand for a reason why this desire is the one that agent chooses might possibly never come to an end, since it is not satisfactorily terminated by the explanation “that’s just how people think.” On the other hand, it might be that brute choice, or a reasonless endorsement, is the way objective-reasons theorists account for particular actions. But then we are left with no explanation why agents would ever act one way rather than another.

While I have here discussed only Searle and Korsgaard explicitly, my point generalizes to many objective-reasons theorists. There is a clouding illusion in all these accounts of action explanation. Their method is to imagine an action, and to identify one of its antecedents that would justify it and serve as its reason—some fact in the world, or some free choice of the agent based on some fact, or on some endorsed desire, or whatever. If the imagined action is Lyle’s taking his umbrella, for example, then his reason for acting this way is the fact of a forecast of rain, or it is his choice based on that fact, or it is his endorsement of his desire not to get wet.

The problem with this explanatory method is that it is just as easy to imagine the agent’s not doing the relevant action while the antecedent circumstances remain the same. There is no contradiction in someone’s not doing what the circumstances of action would justify. It happens every day. In response to this problem, the objective-reasons theorist will want to say that if Lyle decides not to take his umbrella when rain is forecasted, then some other fact about the world must explain this—or he must have freely chosen based on some other fact, or he must have endorsed another desire that competes with his desire not to get wet. But responses like these will not help. Because if Lyle would act for a reason whether he takes his umbrella or not, then which fact of the world he would act on, or which fact would be the basis for his choice, or which desire he would endorse, is arbitrary. None of Lyle’s objective reasons is any more compelling than another. Consider: when we begin by imagining Lyle’s acting, we more or less confidently project some objective reason on which he would act. But when we consider the circumstances of action from his point of view, where any reason would be as good as another, it is hard to imagine his acting at all. This makes him a “lump.”

Objective-reasons theorists may attempt to save Lyle from the debilitating inertia of lump-hood by simply reapplying their method. It may be said that since he can obviously get himself to act on one fact of the world rather than another, some more distant antecedent of his act must serve as a more prior reason. For example, he might have earlier endorsed a principle that favors endorsing one desire, or type of desire, when it conflicts with others. Korsgaard explains how it works for agents like Lyle as follows: “you have a desire to do both A and B, and they are incompatible. You have some principle that favors A over B, so you exercise this principle and you choose to do A…. You regard the choice as yours, as the product of your own activity, because you regard the principle of choice as expressive, or representative, of yourself” (2009, p. 75).

Call the principle of choice invoked here: α. Korsgaard’s advice to Lyle is unhelpful simply because, in the same circumstances, we can imagine his having the alternative principle of choice, which favors B over A. Call it β. Lyle himself can imagine his having β rather than α. So what could make his having α a more compelling expression of himself than his having β? And even if Lyle had endorsed α earlier, what would require him to stick with it now, when facing a choice between A and B? He has no way to answer these questions. Or at least he has no way to answer them that does not raise similar questions again, at even higher levels. So Lyle must remain a lump. Because of all the external facts and internal desires present in his circumstances, his choosing whether to take his umbrella or leave it depends upon his finding satisfactory answers to a series of unending questions.

On the Humean view, which presents desires as reasons for action, the decisive force that saves Lyle from lump-hood resides in the desires themselves. But it is not that in the circumstances of action an agent like Lyle must survey the inventory of his desires and gauge their relative strengths, in order to choose to act on his strongest desire. There are several problems with this idea. It would present another objective-reasons story, with desire-strengths being the relevant facts—psychological facts, in this case. Thus, it would leave Lyle with the problem of needing a reason to act on his strongest desire, rather than the second-strongest, or third, and so on. Moreover, in too many cases it is simply impossible to determine how to rank one’s desires by strength. The only plausible way to determine their relative strengths is to find out which desires are effective in action. In other words, Lyle can know that his desire to take his umbrella is stronger than his desire to leave it only after the fact, only once he gets on his way with or without it (barring forgetfulness!).

It is not that the Humean theory of practical reasons leaves no room for reflective deliberation, as I have shown. Humean agents act on their strongest desires every time, but their beliefs about their actions’ desirable and undesirable consequences, and their conclusions about how effective they are likely to be, tend to regulate their desires’ relative strengths in the circumstances of action. It can sometimes be hard for a Humean agent to decide how to act. For all things considered, the reasons for taking an umbrella and for not taking one can sometimes be equally compelling. But that is not usually the case, fortunately. Usually, one desire prevails, and the deliberation ends. The choice is made.

But if Humean agents always act on their strongest reasons (desires), do they always act on their best reasons? Sadly, as I have suggested, no. Lyle’s reasoning about his taking the umbrella, in the circumstances, may lead him to the conclusion that he had better take it, just in case. But then he may not take it because he finds it cumbersome. As Humeans see it, desire strength, the compellingness of an agent’s reasons, may not be tightly regulated by belief. There can be slippage. Desire can even influence belief, as is commonly supposed.

4 Some recent theories of practical rationality

Some recent accounts of practical reasoning offer different strategies for understanding motivation and normative reasons and for dealing with the leading objection to Humeanism. These views are not Humean, however; they take that normative reasons are in the world. Yet, each offers an alternative to some of the traditional objective reasons theories that I have discussed. I briefly address four here.

Alvarez (2010) has argued that reasons are facts, but she also alleges that desire can sometimes play a role in producing action. On her approach, we reference facts to answer different kinds of questions about an action: why an agent did it (explanatory reasons), the “light in which” the agent did it (motivating reasons), and what made it right (justifying reasons) (pp. 33–37). Desires are sometimes thought of as psychological states and sometimes as goals, she argues, and they are not the right sorts of things to be reasons, but they are sometimes what moves us. How? On Alvarez’s view, a motivating reason is the sort of thing an agent regards as making a prospective course of action good or right, and a motive can be a factor that determines whether certain facts that agent is aware of count as a motivating reason (pp. 185–197). Greed (excessive desire for wealth) might be a motive that determines that the fact a certain investment is likely to yield a huge return is a motivating reason for me to make that investment. If the motive of greed were absent, that fact might not serve as a motivating reason for me.Footnote 16 Hieronymi (2011) offers an account of practical reasons that explains action from the observer’s point of view in terms of facts, but also exhibits the reasons the agent had (if there were some) for doing the action (p. 421). She and other philosophers have noted that regarding as reasons considerations counting in favor of actions or attitudes results in “the wrong kind of reasons” problem.Footnote 17 The problem, which I noted earlier, is illustrated when persons have reasons to adopt a valuing attitude toward an object because they will be paid or favored if they do so. Thus, they have a reason to pursue the object, but that reason has nothing to do with the value of the object itself. Hieronymi suggests instead that we understand reasons “as items in pieces of reasoning,” where reasoning leads to a conclusion. This forces us to consider the question on which a reason bears, rather than allowing that reasons relate directly to action. In this way, we distinguish reasons in terms of the kind of questions to which they pertain: facts that bear positively on the question what to do are normative reasons, and those that explain an action or state of affairs are explanatory reasons. What an agent treats as the considerations she appeals to in order to decide what she ought to do are her operative reasons (p. 422). So, when we explain why Claudia turned off the light, we do it by reference to what she takes to be considerations that settled for her whether to act, that is, by reference to her normative reasons.

While both of these views have something in common with Humeanism, perhaps surprisingly, I think they have more in common with Hume’s view itself. Hume would agree with Alvarez that a motive, what he classifies as a passion for something, makes certain pieces of information relevant to action. While Hume does not think (as contemporary Humeans do) that “rational” is an evaluation of action (rather action is evaluated as virtuous or vicious on the basis of sentiment), he does argue that reason plays a role in the production of action. It does so insofar as it allows us to figure out means to ends or to discover what objects fit the description of the features we desire (T 2.3.3.3). Furthermore, Hume maintains that the function of reason is to lead us logically to conclusions; those conclusions can be relevant to action, although they do not count as reasons for action, but rather as reasons for believing a conclusion. So, Hieronymi’s idea that reasons are regarded as items in pieces of reasoning that prompt us to ask on what concern a reason bears approximates what I think Hume had in mind when he discussed the function of reason and its relation to action. The difference between their views is that, although the items in a line of reasoning for Hume lead to a conclusion that might bear on action, conclusions alone never give us norms or inform us about the value of ends.Footnote 18

Mantel (2018) rejects Dancy’s allegation that a plausible theory of action must be able to identify motivating reasons with justifying reasons. Her project offers a narrative of the person who manifests a competence to be moved to do what the relevant normative reasons indicate. On her view, motivating reasons are the contents of psychological states that correlate with (but are not identified with) justifying reasons in the person who acts the way a virtuous person would. Normative reasons consist in the facts that make our beliefs true, but our deliberation about the facts prior to acting is more detailed and “fine-grained” than the facts themselves. Thus, normative reasons and motivating reasons belong to different ontological categories (p. 112). Mantel maintains that her “Normative Competence Account” allows us to combine “the worldly picture” of normative reasons with the psychology of traditional action theory. “It offers a naturalistic picture of how our mind interacts with normatively relevant features of the world such as to deliver normatively appropriate action” (p. 177).

Lord’s (2018) theory is a “Reasons Responsiveness” theory of rationality: that a rational agent responds to the facts he or she is in a position to know and to use as reasons for the actions they support (pp. 91–93). He agrees, however, with objective-reasons theorists that facts constitute normative reasons. An agent’s true beliefs allow the agent to act from normative reasons he possesses. In the case of acting on false beliefs, the agent performs the action from “a motivating reason,” since he does not possess normative reasons with respect to that particular action. Thus, a motivating reason r is described this way: “what is to ϕ for a consideration r is to ϕ in virtue of the fact that one conceives of r as a normative reason to ϕ” (p. 180). Lord’s is a disjunctive view, in which “everyone reacts for reasons even though many agents who react for normative reasons do not react for motivating reasons” (pp. 178–179). There are many more pieces to this theory—for instance, how to explain that we are acting rationally even if we are all subject to the illusions of a Cartesian evil genius (pp. 184–208)—but my concerns here are different.

All four of these theories of practical rationality accommodate a distinction between motivating and justifying reasons (which Dancy abolishes), and this makes them more plausible than traditional views. None endorses the Humean thesis of desire-based normative reasons. So why is the Humean view preferable? The Humean view is still preferable, I believe, first, because it offers a univocal theory of practical reasons in terms of desires: explanation of action in terms of desire strength and justification of action in terms of desire priority. Alvarez brings desire (motives) into the picture in only some cases; Lord has a disjunctive view of action for reasons. For Hieronymi and Mantel, reference to facts and our cognitions of them does the explanatory and justificatory work, although in different ways. However, second, as I have argued, explanations that reference desire, rather than cognition of facts, better explain why a person does one action over another. Desires have a theoretically measurable or quantifiable force that cognitions do not. One might imagine that reason generates, after deliberation, motivating force commensurate with the content of a judgment or cognition of what it would be best to do (by producing the strongest motive for doing it). However, such a view would imply an extreme and implausible form of internalism whereby an agent always acts on her best judgment (Radcliffe and McCarty 2018, p. 80).Footnote 19 A more moderate view might say that that the more desirable we judge something to be, the more we should desire it. By this principle, it would be unreasonable to desire something more strongly than the degree of its acknowledged desirability. However, since the issue here is action explanation, rather than justification, and we still lack an adequate explanation of why a person would do one action rather than another.

5 Replies to some additional challenges

I return to the objections to Humeanism posed at the beginning of this discussion. Answers to those challenges follow directly from the position I have defended. For instance, one protest to Humeanism I noted is that Humeanism cannot account for the fact that we decide which reasons to act on. As the above discussion shows, however, the idea that we decide which reasons to act on is not a thesis that even objective-reasons theorists can explain coherently. Second, as I mentioned, many non-Humeans (Korsgaard, Raz, Quinn, Dancy) have argued that we have no reason to act on a desire to achieve an end if we have no reason that justifies the end. However, on the view I have defended, all desires give us reasons to pursue their objects, but some of these reasons are better than others. Our second-order desires experienced from a reflective stance negotiate the conflicts between first-order desires and determine which reasons are better for us to pursue. And depending on the version of objective-reasons theory that one holds, attempting to give reasons for the ends results in an infinite regress.Footnote 20

Third, Schueler argues that when one acts on a desire, it is not the desire but one’s belief or judgment about the desire that does the work in explaining action. This point is an objection to the Humean theory as a theory of explanatory or motivating reasons. Schueler considers an act, (A):

I intentionally opened the candy jar, took out some chocolates, and put them into my mouth (p. 176).

And then he writes: “… when I deliberate about whether to perform the action described by (A), it is not the fact that I want to eat some chocolates but rather my judgment that I want to eat some chocolates that comes into deliberation. My deliberation might proceed in exactly the same way even if my judgment that I wanted to eat some chocolates was simply mistaken (unlikely as it might be that one could make such a mistake)” (p. 177). The Humean can agree that deliberation over (A) involves beliefs about the desire to eat chocolate, plus beliefs about competing desires the agent possesses, in addition to beliefs about the consequences of acting on the desire to eat chocolate and on the consequences of acting on competing desires. When the agent forms a second-order desire about the various first-order desires—perhaps a second-order desire that she not desire to eat chocolate just now—there is no guarantee that the second-order desire will have an effect on the strength of the first-order desire.

So, while deliberation on a Humean view can sort out from among various desires’ ends what one values the most, the action that follows is not necessarily reflective of what one values most. It is not reference to beliefs about one’s desires that explain the action, even though they are considered in deliberation. Explanatory reasons must reference what the agent in fact most strongly desires. And even if an agent is mistaken in her judgement about what she desires and so her deliberation is based on false presumptions, in the end it is still the case that her action can be explained only by reference to her strongest desire. If we can be mistaken in our judgments about what we desire, then we can be unaware of what we most strongly desire as well.

A fourth criticism of Humeanism is that doing something in response to a state of affairs counts as doing it for a reason. Thus, doing an action for a reason requires no reference to beliefs and desires unless one is already committed to such a theory in the first place. So Humeanism about reasons begs the question (Bittner 2001; Dancy 2018). This point follows straightforwardly from the leading objection that on the Humean view, an agent can never act on justifying reasons. Of course, the Humean response is to reject the thesis that reasons consist in states of affairs. If my normative reasons derive from my second-order desires and from my beliefs about the world (about how to achieve my ends), then I can be moved by a reason that is justified. I will be moved by such a reason when: (1) my strongest first-order desire is the same as what I desire to desire and (2) my beliefs about means to the end are properly formed. If I am pleased from the second-order perspective over my desire to eat vegetables (I am pleased because I know they are part of a healthy diet), then my eating vegetables is justified by my reasons. While it is not the case that my second-order desire is itself moving me, it does make sense to say that I am acting with good reason. Second-order desires also might sometimes move me, not because they are higher-order but because they might be strongest among the desires I possess at the time. But because second-order desires are desires to desire, the sorts of actions that they motivate are those that make first-order desires stronger, whatever they might be (for instance, doing, as practice, the action the approved first-order desire indicates; seeking professional help; putting myself in conducive circumstances, etc.). It is true that the Humean account of reasons that I offer is not necessarily an account of doing an action because it is right. Neither is Dancy’s account of objective reasons one of doing an action because it is right. In his narrative, an agent acts in response to the world. That the world is the way it is makes certain actions justified and others not. On the other hand, I have suggested in Sect. 2 above a way that a Humean can countenance talk of doing the right thing.

Furthermore, one wrinkle in Dancy’s view is what we are to say when the agent has false beliefs. On his view, all reasons are normative, since all refer to the world’s being a certain way; thus, even explanation of action is not in terms of agent beliefs. If the way the agent sees the world is not the way it is, then the explanation of action, Dancy claims, is in terms of the considerations in light of which the agent acted, even though they are “non-factive.” (2000, pp. 131–137). This is mysterious, and Dancy himself puts it in terms of acting for a reason that is no reason: “A reason is just a consideration that speaks in favour of an action (or belief). Its being the reason it is consists in its so speaking—its supporting this action rather than that one. So, a reason that is no reason still exists, in a sense, as the consideration or feature that the agent wrongly took to support acting in the way he proposed, though the supposed reason (or support it gives to acting in the way he supposed) does not” (p. 144).

In his more recent book, he writes,

… if people reason from considerations that are not the case, it is not as if they are reasoning from nothing—from a kind of blank. It is for them as if those things were so, and they are reasoning in the light of that mistaken conception (that is, in the light of the way they mistakenly conceive things to be). So certainly it is not all a disaster. They have made one mistake, yes, but only one, and having made that mistake they then continue along pretty well, one might say. Their deliberation, given their starting point, may even be impeccable. But still it is all built on nothing. (2018, p. 41)

But how can it be the case that, for people who reason from considerations that are not the case, it is not as if they are reasoning from nothing and yet their deliberation is built on nothing? Mele (2007) attempts to make sense of the former passage, but cannot; he proposes, rather that we accept a view on which there are normative reasons of the sort Dancy describes and explanatory reasons described in terms of an agent’s psychological states, which Dancy rejects. While I am not here accepting Mele’s proposal, I only want to suggest that Dancy’s presentation of objective reasons is problematic in ways that he cannot easily address.

Finally, I have argued that Humeanism can account for both explanation and justification of action, while objective-reasons views have a difficult time accounting for explanation of action. It is a fair question to ask, however, whether we can realistically expect any theory to offer an explanation why certain reasons are being acted upon and others are not. In fact, it is not clear that the psychologist’s statement that one desire was stronger than another is better, if we cannot predict what reason an agent will act on.Footnote 21 In reply, Humeans can say the following. First, we do accept references to causes as explanations in science, even though we never arrive at an “ultimate” cause, which would take us into metaphysics. Just as science stops at the bounds of the natural world, so too, Humean explanations of action stop with the mental states that are causes. While we cannot predict based on how a motive feels what an agent will do, we can make some predictions based on the agent’s track record of acting on certain sorts of desires in the past. If we take seriously the idea of character development and strengthening motives by habituation, then it is plausible to think that we are able to predict what persons with fairly stable character will do in many situations (barring character skepticism, about which there is a huge controversy). Therefore, the Humean explanation of action does no better or worse than other empirical explanations of phenomena.

6 Conclusion

The prospects for Humeanism about justifying reasons are bright. However, can Humean agents act for reasons that justify? Can one truthfully say, “I’m not doing it because I want to, but because I ought to”? Can one say, “I have no desire to do it; I’m doing it because it is right”? As I noted earlier in this essay, Hume’s explanation of acting from duty was in terms of acting from a reaction to our lack of certain motives. He thought agents who lack a certain virtue could do the action a person with that virtue would do by being motivated by a feeling of self-disapproval, perhaps akin to self-disgust or guilt. They would not however, be acting from the thought that “this is the right thing to do;” rather, the motive is still a conation and not a cognition. On the Humean perspective that regards all motives as desire-belief combinations and reasons as comparative, the explanation is a little different. When the agent reflects on her desires and approves of one over another, she has better reason to act on the former than on the latter. Sometimes the reflective process might strengthen the approved second-order desire, although there is no guarantee that this will happen. When it does happen, however, it makes sense to say that the agent is performing the action, taking the means to the end, because it is the right thing to do; the action is nonetheless still a product of desire.

To say, for instance, that I helped with the benefit event, even though I had no desire to, is a way of speaking about the phenomenon, the feeling, of desiring. On a Humean view, no one acts without a desire for the end one pursues. However, sometimes we act on desires that do not feel “exciting” to us; nonetheless, such desires are the causally strongest ones among those in our psyches and so do “excite” us to act.Footnote 22 So, to us, it feels as though we had no desire to do the action we thought we ought to do. On the Humean theory of practical reasons, expressions like “I did my duty, even though I didn’t want to” additionally express an agent’s recognition of the difference between reasons or desires that are merely personal, and those that arise from reflective thinking on human nature and social life, from a general point of view. It might be that such expressions are never true in a precise sense. But nothing in the theory prevents agents from acting on what Humeans recognize as the better desires of human nature.