Because I Believe It’s the Right Thing to Do

Abstract

Our beliefs about which actions we ought to perform clearly have an effect on what we do. But so-called “Humean” theories—holding that all motivation has its source in desire—insist on connecting such beliefs with an antecedent motive. Rationalists, on the other hand, allow normative beliefs a more independent role. I argue in favor of the rationalist view in two stages. First, I show that the Humean theory rules out some of the ways we ordinarily explain actions. This shifts the burden of proof onto Humeans to motivate their more restrictive, revisionary account. Second, I show that they are unlikely to discharge this burden because the key arguments in favor of the Humean theory fail. I focus on some of the most potent and most recent lines of argument, which appeal to either parsimony, the teleological nature of motivation, or the structure of practical reasoning.

It appears evident… that those actions only can truly be called virtuous, and deserving of moral approbation, which the agent believed to be right, and to which he was influenced, more or less, by that belief.

–Thomas Reid

Former Governor of the state of Florida, Charlie Crist, recently proposed to have Jim Morrison pardoned for some allegedly lewd behavior in 1969 at a concert in Miami. Interestingly, Crist said “I’ve decided to do it, for the pure and simple reason that I just think it’s the right thing to do.”¹ Certainly normative beliefs such as this sometimes have an effect on what we do, but their precise role is controversial. So-called “Humean” theories, which insist that all motivation has its source in desire, attempt to connect such beliefs with an antecedent motive.² Bernard Williams (1979/1981), for example, is commonly read as denying that practical reasoning can generate a new desire when “there is no motivation for the agent to deliberate from” (p. 109). Such Humean views are abundant.³ On the other hand, those often labeled “rationalists” eschew such commitments, permitting normative beliefs to sometimes play a more independent role in rational motivation.⁴ This alternative view allows us to take Crist’s claim at face value, since the relevant explanation of his action could then terminate in the belief he cites.

Achieving progress on such a debate is perhaps difficult from the armchair given the decidedly empirical aspects of the competing theories. But it is not impossible.⁵ We can do so, I shall argue, by first providing an initial case for rationalism, which appeals to the ubiquity of ordinary explanations of each other’s rational actions that do not appeal to the antecedent desires posited by the Humean theory. This shifts the burden of proof onto Humeans to motivate their more restrictive, revisionary explanations. Yet they are unlikely to discharge this burden, because the key a priori arguments for the Humean theory—i.e. for believing there must be some antecedent desire—fail. I shall focus on three such arguments for Humeanism, which attempt to capitalize on either parsimony, the teleology of motivation, or the structure of practical reasoning. Without these key lines of support, Humeans are left with no substantial tools for discharging their burden of proof, leaving us with reason to be rationalists.

1 Humeanism vs. Rationalism

Before providing a case for rationalism, we must develop a perspicuous account of it and the competing view. An important starting point is an account of the kinds of actions that are of interest. At least in humans, intentional actions tend to make sense; they are in some sense rational or at least intelligible. The Humeanism-rationalism debate concerns the causal or explanatory structure of such acts. In particular, we are concerned to account for actions with a so-called rationalizing explanation, which makes sense of the agent’s action by revealing the “favourable light in which the agent saw his projected action” (McDowell 1978/1998, p. 79). As Davidson famously puts it, one’s reason for acting “rationalizes an action only if it leads us to see something the agent saw, or thought he saw, in his action—some feature, consequence, or aspect of the action the agent wanted, desired, prized, held dear, thought dutiful, beneficial, obligatory, or agreeable” (1963/1980, p. 3).

Suppose, for example, that a man turns on radios whenever he believes they are off (cf. Quinn 1993/1995). At some point he intentionally turns on a particular radio and so he had, prior to the time of action, a desire to turn it on. This man’s act is not rationalized by his relevant mental states, since together they do not reveal any aspect of the action that he thought favorable. He simply believed the radio was off and so wanted to turn it on. Of course, to make sense of such an action, we might assume, say, that he is averse to silence (i.e. that he desires some noise); but we are imagining that is not so in this case.⁶

Now consider the following scenario, adapted from Stephen Darwall (1983, p. 39).

Roberta’s Moral Belief

Roberta grows up comfortably in a small town, which presents her with a congenial view of the world and her place in it. On going to a university she sees a film that vividly presents the plight of textile workers in the southern United States. Roberta is shocked and dismayed; she comes to believe she has a moral obligation to help such causes. Because of this belief, she decides after the film to donate a few hours a week to promote a boycott of the goods of one company that has been particularly flagrant in its illegal attempts to destroy the union.

How should we explain such a case? In particular, we might wonder whether Roberta’s belief could produce and rationalize the motivation to help without serving or furthering some antecedent desire. Humeans, unlike rationalists, deny this possibility since they endorse what we might call an “instrumentalist” account of the structure of (normal) motivation.

Some of Thomas Nagel’s (1970, p. 29) terminology is helpful when addressing such connections between beliefs and desires. Let us say that motivated desires are in some sense grounded in “reason” by figuring in a rationalizing explanation (cf. Wallace 1990/2006, pp. 23–4). For example, my desire to take some acetaminophen is “motivated” if I acquire it because I believe it will satisfy my desire to relieve my headache. On the other hand, unmotivated desires are motivational states that do not figure in a rationalizing explanation. For example, Donald’s sudden yen to touch someone’s elbow, something he just finds himself with, is unmotivated. Similarly, familiar “deviant” mental causes are not motivated, since they explain without rationalizing. The idea is not that these desires fail to be generated or sustained by conscious processes of thought. Rather, unmotivated desires are “appetites” or “passions” in one sense of those terms.⁷

So rationalists can agree with Humeans that desires for what one believes to be a means to one’s ends are motivated. The key difference is that rationalists believe some motivated desires can be grounded in more than such broadly instrumental or means-end reasoning. For example, one might simply believe it’s best to return a wallet and because of that ultimately desire to do so. But Humeans maintain that if there are any actions admitting of a rationalizing explanation, they are those generated by a process of broadly instrumental reasoning, in which an antecedent desire is served or furthered by any subsequent desires (cf. Williams 1979/1981).

We can further clarify the issue by employing the typical distinction between ultimate and instrumental desires, already alluded to in the previous paragraph.⁸ As we’ve seen, the issue between Humeans and rationalists is generally this: Does all motivation ultimately have unmotivated desires at its source? Even more concisely, the question is: Are all ultimate desires unmotivated desires? Humeans answer affirmatively, rationalists negatively. We now have a concise way to represent the claims of each party:

Motivational Humeanism: All ultimate desires are unmotivated.

Motivational Rationalism: Some ultimate desires are motivated.

Each thesis makes a claim about the causal and rationalizing profile of certain mental states and the explanations in which they figure. One can see then that there are two main routes to Humeanism. Some theorists focus on the more conceptual claim that an explanation only counts as rationalizing if it reveals how the action promotes one of the agent’s ultimate desires (e.g. Lenman 1996). Others, however, emphasize the more empirical part of Humeanism, insisting that rationalizing explanations always in fact have this causal-explanatory structure (e.g. Sinhababu 2009).⁹

Returning to Roberta, motivational Humeans can explain her desire to help the workers by appealing in part to her belief that she should. (Recall the case stipulates that the moral belief plays a causal role.) But, if we are to provide a rationalizing explanation of Roberta’s action, they must also then appeal to some antecedent desire. Perhaps the most plausible option is to attribute to her an ultimate desire to do what she should. But given such a desire, Humeans must hold that we cannot provide a rationalizing explanation of it (i.e. it’s not a motivated desire). Her having that desire is explained by something else, such as a moving experience, habituation, or being hit over the head with a hammer. Rationalists, on the other hand, needn’t appeal to such an antecedent desire. They can explain and rationalize her desire to help (thus making it an ultimate and motivated desire) by appealing to her normative or evaluative belief.¹⁰ On this account, the source of Roberta’s ultimate desire is “pure reason” in some sense, since the ultimate desire is explained and rationalized by her conception of what she should do without appeal to any antecedent desire.

Notice that motivational rationalists admit what some might be content calling a “Humean” belief-desire psychology. In particular, they agree that intentionally A-ing always requires a desire to A, in the broad sense of state whose function is to bring about its content. Moreover, they agree that beliefs alone are not essentially motivational states because they lack the relevant “direction of fit” (see esp. Darwall 1983). So they needn’t believe in so-called “besires” if these are defined as states whose function is both to represent and be efficacious (i.e. have both directions of fit). What rationalists do maintain, however, is that sometimes there is a rationalizing explanation for a person’s action that terminates in a belief rather than a desire. In other words, while beliefs are not motivation-encompassing attitudes, they can be directly motivation-producing—to use some of Alfred Mele’s helpful terminology (2003, ch. 1). This enables the rationalist claim that one can do something ultimately because one believes it’s the right thing to do.¹¹

Of course, non-cognitivists provide a different story. They would object to both explanations simply because they believe there are no moral beliefs for Roberta to have. One’s “moral judgments,” they say, are just expressions of non-cognitive attitudes, which can’t be true or false. For expository purposes, I will often assume the truth of moral cognitivism—the view that moral judgments are genuinely cognitive mental states with truth-apt, representational contents. So the brand of motivational Humeanism I primarily seek to undermine maintains a cognitivist view. The dispute to highlight here is over the causal-rationalizing powers of normative beliefs, not whether they exist. Furthermore, our focus will be on the broader category of normative or evaluative judgments (e.g. with contents involving the concepts of ought, reasons, should). So the assumption will be an even broader view we might call “normative cognitivism.” However, as we will see, non-cognitivism doesn’t fare much better anyhow.¹²

2 Revisionary Explanations

Focusing on the ordinary explanations of each other’s actions, we can develop a presumptive case for motivational rationalism. Recall Roberta who, upon watching a moving film about the plight of some textile workers, comes to believe she has a moral obligation to help such causes. This moral belief then causes Roberta to donate some of her time to help. Humeans must connect this belief with an antecedent desire, such as the desire to do whatever is right (provided they want to count her action as rationalized). Darwall says: “This need not be what happened” (1983, p. 40).

Neil Sinhabau (2009) has recently responded to Darwall by saying that Roberta does in fact have an antecedent desire. Given that Roberta is “acting to relieve suffering,” he posits an antecedent “desire that others not suffer” (pp. 486–7). This strategy appears to involve inferring Roberta’s mental states and motivational structure from how the case is described. For example, he maintains that “Roberta’s shock and dismay serve as evidence for the presence of an antecedent desire” (p. 485).

While Darwall does provide various details about the case, at least one way of appealing to cases like Roberta’s involves more stipulation than Sinhababu allows. In our version of the case, we have simply stipulated the following. Roberta has a moral belief, namely that she ought to help, and that belief plays some causal role in bringing about her act of helping. This much stipulation does not beg the question at issue since Humeans can say that, while the belief plays a causal role by generating a subsequent desire to help, an antecedent desire is the ultimate cause. I take it that Darwall’s basic point is that there are in fact such cases in which there is no antecedent desire, so the Humean owes us an explanation of why these cases are impossible. In fact, this is one way to read Scanlon’s case of giving a friend “unwelcome news” (1998, p. 39). Tim does not want to deliver the news but believes he ought to do so. One way of at least appropriating such as case is to imagine the same stipulations as before: Tim has a moral belief (that he ought to deliver the news), and it plays a causal role in bringing about the corresponding action. Moreover, this appears to be a rationalizing explanation since believing something is right plausibly reveals something the agent thought favorable about the act. Again, the problem for Humeans is that there seem to be such cases in which this is the extent of the story, so the burden of proof is on them to tell us why we should believe otherwise.

To illustrate the worry, we can consider a similar argument Michael Smith levels against an ambitious form of motivational externalism. This is roughly the view that one can fail to be motivated to act in accordance with one’s moral judgment yet remain practically rational.¹³ Smith (1996) has us consider a case of change in motivation based on moral beliefs. It involves a friend converting to utilitarianism after reading Peter Singer’s work and then later switching back to a view embracing partiality. Suppose this friend’s actions and motivations change according to his moral beliefs. For example, he helps strangers just as much as kin at one point, but then later focuses his efforts more on his inner circle. Smith contends that motivational externalists must explain the changes by appealing to an ultimate desire to do whatever it is that he thinks is right. And this is meant to be read “de dicto” rather than “de re.”

Smith doesn’t fully explain this distinction, but the idea is simply that the different readings affect what features in the content of the desire one is attributing to the subject. On a de dicto reading of “Roberta wants to do what she thinks is right,” for example, the “rightness” is part of the content of the desire (Roberta wants to: do whatever she thinks is right). On the de re reading, the agent simply wants to perform the action (Roberta desires to: help) and separately believes that action is right. Here the “rightness” is only part of the content of the belief. The following provides a more schematic representation of the kinds of explanations of an action that correspond to each reading (parentheses indicate what is in the content of the mental state and the arrow represents causation):

De re: Believe (A-ing is right) → Ultimate Desire (to A) → A

De dicto: Ultimate Desire (to do whatever is right) + Believe (A-ing is right) → Instrumental Desire (to A) → A

So, according to Smith, externalists must say that the agent has an antecedent desire (de dicto) to do whatever it is that ends up being right, rather than simply desiring (de re) to perform a particular action, A, which the agent believes is right. On the de dicto picture, Smith objects that “we must redescribe familiar psychological processes in ways that depart radically from the descriptions that we would ordinarily give of them [i.e. descriptions involving only a de re desire]” (p. 181).¹⁴

A similar charge can be made against Humeanism. When confronted with cases like Roberta’s Moral Belief, in which a normative belief plays a causal (and apparently rationalizing) role in the production of action, Humeans are forced to attribute an antecedent desire to the agent. But which? If the subsequent mental states are to serve or further the antecedent desire, then it must be something like the (de dicto) desire to do whatever is right. Nick Zangwill rather explicitly embraces this: “The motivating desire is the desire to do the morally preferable thing—or perhaps to do the right thing” (2003, p. 144). Humeans could posit a different antecedent desire (cf. Finlay 2007), but the problem remains: they are forced to come up with some antecedent desire or other. Rationalists, on the other hand, can attribute merely the de re desire to perform the action, which may be generated because the agent believes it’s right. As in Crist’s case, rationalists can provide the ordinary (rationalizing) explanation of Roberta’s action, which appeals only to her moral belief generating a desire to act in accordance with it. They needn’t redescribe or read into the case to conform with a theory that requires an antecedent desire.

Humeans might object that mere appeals to common-sense psychological explanations are rather frail. But the argument does not rest on a “mere intuition” devoid of any other substance or support. Ordinary appeals simply to normative beliefs are quite ubiquitous.¹⁵ Roberta’s story is fictional, but recall Crist’s perfectly quotidian description of the reason he proposed the pardon of Morrison. He did it simply because he “thought it was the right thing to do.” Rather often we attribute only a belief about what’s right as the ultimate cause of one’s action. And we do this, not just for ourselves, but for others. Jamie McCabe, son of the first Australian to successfully sue a tobacco company for her lung cancer, said of his late mother, Rolah: “She did it because she believed it was the right thing to do.”¹⁶ Cases like this abound in ordinary discourse, and vastly outnumber explanations in purely conative terms (e.g. “because he wanted to do the right thing”). If doubtful, a simple search the Internet yields ample confirmation.¹⁷

Of course, in all these cases, Humeans can explain away cognitive attributions as either erroneous or elliptical (i.e. as an abbreviated way of attributing an antecedent desire). But the burden, I submit, is on them to support such alternative readings, given the methodological principle that, all else being equal, natural discourse should be taken literally. As Tyler Burge puts it, in the context of a different debate, “unless there are clear reasons for construing [natural] discourse as ambiguous, elliptical, or involving special idioms, we should not so construe it” (1979/2007, p. 116). Otherwise, if taken at face value, Humeans must treat this ubiquitous practice as involving explanations that don’t rationalize anyone’s actions. This is implausible on its face since normative beliefs do seem to reveal what the agent thought favorable about the action. One could argue otherwise on theoretical grounds, but we’ll turn to such attempts later and see that they are unconvincing.

Moreover, we can uncover some reasons for why we might find it plausible that we often attribute merely the normative belief. Consider the antecedent desire Humeans attribute to Roberta and similar agents. How, on this view, did she come to have this desire to do whatever is right? Are we all either born with such a desire or lacking it, or with its various strengths? Humeans could of course appeal to other psychological states and processes one might have throughout one’s lifetime, such as association or certain experiences that generate the antecedent desire (Finlay 2007, n. 12). But one thing Humeans cannot cite is a process of reasoning, or anything that counts as a rationalizing explanation. Yet these seem quite relevant to the existence of a desire with normative content (cf. Barry 2010, sect. 4). Humeans can cite beliefs as responsible for such antecedent desires, but these cannot count as rationalizing explanations or causes. Positing an antecedent desire, then, seems insufficiently motivated on their account. Of course, some theorists argue that such standing desires, such as a desire to act in accordance with reasons or a desire to do what makes sense, are constitutive of being a rational agent (e.g. Velleman 1992). But these are rather contentious claims, which would further constrain the Humean theory.

Of course, motivational rationalist likewise must hold that some people simply transition from a normative belief to the relevant desire while others don’t. But this only amounts to saying that some people are better constituted than others and that we needn’t be able to explain this by appealing to an antecedent desire. Instead, rationalists can appeal to a mere disposition to transition from the normative belief to the subsequent desire, assuming mere dispositions, lacking any content, are not desires in the relevant sense (for defense of this assumption, see Dreier 1997, p. 94). This yields some common ground: some psychological constitution or other—having some kind of desire, belief, or even bare disposition—is required for rational agency. Rationalists, such as Korsgaard and Smith, can hold, for example, that a rational agent will be disposed to desire to A upon believing that A is right (or best, etc.). More Aristotelian rationalists might prefer to describe such a person as virtuous, but what unites these theorists against Humeanis is that they do not posit an antecedent desire with the content do whatever is right, which is then promoted by the subsequent belief that A is right and the desire to A. That is a broadly instrumental explanation, which rationalists need not provide, since a mere disposition is not a state that specifies a goal that is served or furthered by the transition from the normative belief to the relevant desire. A substantive dispute remains, since rationalists provide an account on which “reason” is not a slave to the “passions.” One can, for example, be motivated to do what’s right by coming to believe that it is, without appealing to something one already wants (for further discussion, see May ms).

Our ordinary tendency to avoid explaining actions in terms of antecedent desires might also be due to our implicit conception of certain forms of rational motivation. There is something of “rational worth,” as we might call it, in one’s desires being responsive simply to one’s recognition of what one should do, rather than some antecedent desire.¹⁸ We needn’t insist that all rational agents believe what they’re doing is right or lack the desire to do whatever is right. The idea is simply that our ordinary explanations of one another’s actions leave room for acting simply from recognition of one’s duty. Kant was at least right that this is something we value. Of course, this largely speculates about two issues. First, this might not be part of our ordinary conception of rational motivation. Second, our ordinary thinking might not stand up to philosophical scrutiny. Exploring these issues is beyond the scope of this paper. The point to register here is merely that the appeal to common-sense psychological explanation is not obviously flawed on its face.

Nevertheless, one might continue to press the worry: What kind of hold does an argument have when it is based in common-sense or intuitive explanations of empirical phenomena? To answer this question, it will prove instructive to consider a related argument. Many rationalists implicitly assume that their conceptions of motivation are the more intuitive or pre-theoretically plausible view. Russ Shafer-Landau (2003), for example, details this sort of presumptive case against Humeanism by appealing primarily to the “phenomenology of motivational experience.” Pre-theoretically, he claims, “most of us would find it plausible to suppose that some evaluative beliefs… can motivate all by themselves” (p. 123). He gestures toward the kind of intuitive picture of temptation according to which what is battling out for motivational influence is a desire for what is tempting us and a belief that one ought to resist. (Compare Hume’s characterization of the dominant view in his time of “the combat of passion and reason” in the Treatise [1739–40/2000], 2.3.3.1.) Shafer-Landau helpfully compares the situation of Humeanism here to psychological egoism—the thesis that all of our ultimate desires are self-interested. Both views are subject to a kind of burden-shifting argument, saddled with the task of explaining away the more intuitive and non-revisionary view.

Although Shafer-Landau recognizes that such phenomenological considerations are limited in force, we should pause to examine their value. As Jay Wallace (1990/2006) points out, we must tread carefully when considering phenomenological arguments for either the Humean or rationalist view. Since the notion of desire in play is broad enough to include mere motivational states of which we are unconscious, “it is unclear how phenomenological or experiential evidence could possibly settle the issue of whether desires always serve as the ultimate source of our motivations” (p. 41). Such worries are only amplified when we consider that introspection provides rather fallible access to our own mental states.

However, this can be reconciled, leaving the argument against Humeanism intact. First, introspective fallibility is compatible with phenomenological evidence doing minimal epistemic duty, such as merely shifting the burden of proof or tipping the balance in an explanatory tie. Consider again the analogy with psychological egoism. On such a related issue concerning moral motivation, we readily shift the burden of proof onto the egoist because it seems that we can sometimes have genuinely altruistic desires.¹⁹ We wouldn’t want a double-standard applying in the case of Humeanism. Second, despite Shafer-Landau’s label, his arguments, as well as the one developed here, don’t appear to be grounded primarily in phenomenology anyhow. They are meant to support motivational rationalism by providing intuitively better explanations, ones that cohere better, not merely with phenomenology, but with our common folk-psychological explanations. This is something that can be grounded in more than phenomenology, as we saw above.

One might concede this much but argue that armchair reflection and cursory archival searches aren’t a reliable way of gauging what is or is not a common-sense explanation. While there certainly may be better methods for testing what are largely empirical questions, we should keep in mind two points. First, not all empirical questions need more rigorous testing. For example, even though it is compatible with various empirical evidence that, say, the experimenter fabricated the data or made a thoughtless error, we don’t often call for additional empirical tests for this unless there is reason to suspect it. Second, recall that we are only attempting to employ these considerations as a sort of tie-breaker—as involving rather minimal epistemic duty. This bears an apt resemblance to the Humean’s weapon of choice here, Ockham’s razor, which is often used in a similar way.

So far we have been assuming that there are such things as moral, or more broadly, normative beliefs. And it may seem that non-cognitivists can easily avoid this problem of revisionary explanation by denying their existence. But this strategy does not automatically evade the issue. First, one must hold that the judgment about A-ing is identical to (or for some reason entails) the desire to A, which is not constitutive of non-cognitivism. (It is perhaps common to many versions of emotivism, but not necessarily prescriptivism, for instance.) Second, some non-cognitivists only apply their view to morality and do not extend it to all normative judgments. Yet we have construed Humeanism and rationalism as concerning the broader category. And the argument is about ordinary explanations of changes in motivations due to (practical) normative beliefs. So merely moral non-cognitivists—who are not also “normative non-cognitivists,” as we might call them—won’t necessarily get around the problem. They might allow that there are, for example, prudential beliefs with normative content that can generate the corresponding motives. Given the universal claim of motivational Humeanism, such a theory would be a form of motivational rationalism. So it is only a wholesale normative non-cognitivism, and one that identifies the judgment with the desire, that can circumvent the challenge while remaining a form of Humeanism.

3 Why Be Humean?

So far we have seen that motivational rationalism fits nicely with some of our ordinary explanations of action. Humeans, of course, can argue that such cases are misleading on their face, maintaining that a rationalizing explanation must include some antecedent desire that connects with the belief about an action’s rightness. But must there always be some antecedent desire? Supporting such a universal claim that doesn’t seem to enjoy the status of common sense requires argument. The burden is on Humeans to defend their alternative account of the ordinary explanations we’ve identified. Yet the most prominent arguments, as we’ll see, are fraught with problems.

3.1 Parsimony

Most motivational Humeans ultimately employ Ockham’s razor to establish their view. They tend to admit a sort of stalemate: both theories can provide adequate explanations. But they appeal to the famed razor to shift the case in their favor. Sinhababu (2009) has recently employed this strategy in an explicit way:

The Humean theory offers us the attractive promise that a simple explanation invoking only desire-belief pairs for motivation will be sufficient to account for all cases of action. If this promise cannot be kept, we will have reason to go to theories drawing on a more expansive set of explanatory resources—perhaps theories according to which beliefs about our reasons are capable of causing action or generating new motivational forces without any assistance from desire. (p. 466)

Perhaps the same move can be made here: discharge the burden in the name of parsimony. But there are at least three problems with such a Humean strategy.

First, there is antecedent reason to avoid staking one’s account of motivation on simplicity. The history of psychological theory has shown a trend in the proliferation of moving parts, such as types of mental states, processes, or modules. At this point, the value of even seeking to appeal to Ockham’s razor is not immediately obvious, at least given the domain in which it is being employed. Reductive programs might seem appropriate in physics, but psychology and related fields have tended toward introducing new entities and mechanisms, not eliminating them (cf. Haidt and Bjorklund 2008, pp. 205–6). Consider memory as an example (cf. Holton 2009, pp. xii–xiii). Rather than develop a unified conception of memory, psychologists have posited quite distinct kinds with rather different functions (e.g. short-term, long-term, declarative, procedural, episodic, semantic). Of course, a Humean might reply that they are still all memory, with features that unify them under that genus (e.g. they’re all ways of storing information in one’s mind). Similarly, Humeans do not shy away from distinguishing different types of desires (cf. Mele and Sinhababu). However, the general point still holds, which is that we should in advance expect the architecture of our evolved minds to be rather disjointed and modulated rather than simple and elegant. While the razor may still be of value when all else is equal, we might at least bet that its role in psychological theorizing will be limited.

Perhaps, though, we shouldn’t be antecedently dubious of parsimony in moral psychology. Even if so, however, a second problem is that it’s unclear whether the Humean explanation is in fact any more parsimonious.²⁰ The kind of motivational rationalism primarily defended here doesn’t posit a distinct kind of motivational mental state. Given that rationalists can agree that beliefs are not conative states, they needn’t posit so-called “besires” which have both directions of fit. Instead, rationalists hold that there is no need to always appeal to an antecedent desire. They do admit a kind of transition that the Humean does not: a rationalizing explanation involving a transition between only a belief and desire. But roughly the same kind of transition is countenanced by Humeans as well. They after all admit that sometimes we transition, in a rationalizing way, between a belief and desire. They simply think that this only happens when there is a relevant antecedent desire. Undoubtedly, on one way of counting, rationalists posit a certain kind of mental process that Humeans do not. But this process is not radically different in character from any one Humeans recognize. So it is at least quite unclear whether one could legitimately employ Ockham’s razor in favor of Humeanism.

Given these problems with the appeal to parsimony, I submit we follow Hume himself in being wary of that “love of simplicity which has been the source of much false reasoning in philosophy” (Hume 1751/1998, App. 2.6). However, even if parsimony can properly adjudicate between Humeanism and rationalism, it applies only if the competing theories provide adequate explanations of the data. That is, the Humean and the rationalist must provide equally good explanations of intentional action. All else must be equal. If this holds and Humeanism requires positing fewer entities, then there may be some reason to favor it over rationalism. So we cannot rest content with providing an explanation of the data if the Humean explanation is a less plausible one—a point also emphasized by Melissa Barry (2007). As theory is always underdetermined by data, there will always be multiple hypotheses in play. Yet in all areas of inquiry we are content with ignoring the implausible ones. We don’t, for example, take seriously a hypothesis that explains some data by appealing to the intervention of Martians. A Popper-inspired philosopher might suggest that we ignore such hypotheses because they aren’t falsifiable. But suppose this one is. There are many falsifiable hypotheses we don’t bother testing because they are implausible. Similarly, we sometimes opt for certain theories over others, whether in philosophy or psychology, because they are more plausible.

Appealing to implausibility is precisely the sort of argument we have developed above. Sinhababu does seem to suggest that his account of cases like Roberta’s is better than his opponents. But even illuminating all the various aspects of desire that Humeans can employ cannot alone discharge the burden of attributing antecedent desires.

3.2 Teleology

If parsimony can’t meet the burden placed on the Humean’s revisionary explanations, perhaps more substantive considerations can. Arguably the most prominent case for Humeanism builds on Smith’s (1987) well-known teleological argument. In short, Smith relies on the idea that desires are goal-directed states and motivation requires this sort of teleology. But as we’ve seen motivational rationalists, following Nagel and even Smith himself, do not deny such claims (cf. Wallace 1990/2006). They maintain that intentionally A-ing requires a desire to A, but the desire can at least sometimes be generated by a normative belief. However, one might try to extend the argument such that the same old teleological considerations establish motivational Humeanism. This is precisely what James Lenman (1996) attempts to do in support of his Humean “belief-desire theory.”²¹

Lenman rightly takes a view like Nagel’s or Darwall’s as a key competitor. According to Darwall (1983), Roberta’s motivation to help could have been rationally explained by her belief that she ought to, without the help of some antecedent desire (e.g. a standing desire to do what she thinks she should). Lenman’s key strategy, by way of response, is to only admit the conceptual possibility of the causal connection—i.e. that Roberta’s normative belief can causally bring into existence the corresponding desire without the assistance of an antecedent desire. However, he seems to think such transitions from belief to desire are entirely arbitrary, for he insists that it is conceptually impossible for such a description to yield a rationalizing explanation:

My coming to believe that P only begins to provide a rational explanation for my coming to desire that not-P if there is something about P to which I am averse. This is just the teleological argument over again. (1996, p. 294)

But this builds in considerations beyond the original teleological argument (a la Smith). Let us examine this closely.

As Lenman himself puts it, the “central claim” of Smith’s teleological argument is only that “intending to φ, in virtue of being a goal-directed state, involves having a [desire]: so that anyone who intentionally φs necessarily has some desire, presumably at least the desire to φ (under some description)” (p. 292). These considerations alone only establish that performing some action A requires that one desired to A, which rationalists needn’t deny. What Lenman assumes in his extension of the argument, however, is much stronger. He maintains that explaining someone’s action simply in terms of desiring in accordance with one’s normative beliefs does not count as rationalizing. Indeed, on his view, Roberta’s desiring to help independent of an antecedent desire is as rationally (though not causally) inexplicable as desiring to put some parsley on the moon upon believing that there is none there (p. 294)—one of Nagel’s own curious examples (cf. also our radio case, adapted from Quinn). This alone should make us pause. Nagel once said: “A theory of motivation is defective if it renders intelligible behaviour which is not intelligible” (1970, p. 34). Likewise I submit that a theory of motivation is defective if it renders unintelligible behavior which is patently intelligible. Unlike Nagel’s parsley example, believing an action is right clearly reveals the favorable light in which the agent sees the action.

As we can now see, Lenman’s argument goes beyond Smith’s. In doing so it assumes a great deal about the nature of practical reason and reasoning. The crucial assumption is that the following is not alone rational: desiring to A because one thinks one should A. This is what apparently licenses his claim that explaining the relevant desire by citing the corresponding normative belief is not a rationalizing explanation. But why say this? There seem to be two responses in Lenman’s view, but both are problematic.

First, Lenman seems to assume something like instrumentalism about practical reason. That is, he takes it as obvious that the only (normative) reason for doing anything (or having a certain attitude) is that which will satisfy one’s desires. He says: “To reason, theoretically or practically, we must, it is plausible to suppose, accept some such norms, general or specific; and in my present broad characterization of “desire”…, the acceptance of such norms counts as having desires” (p. 296). In other words, Lenman believes it’s a conceptual truth that we are not engaged in reasoning unless we desire to reason in accordance with the relevant norms. So the only kind of transition that he will allow as rational is one in which the generated desire serves or furthers the antecedent one. But there are two problems with this position.

The first problem is that instrumentalism is a rather controversial view that is far from obvious. In fact, it seems Lenman is implementing precisely the dubious strategy Christine Korsgaard (1986/1996) identifies in many motivational Humeans (in her terms “motivational skeptics”). She claims they only seem to deny the motivational force of reason by implicitly denying the existence of any non-instrumental norms of reason. Hidden beyond the skepticism about reason’s motivational powers is a view about the limits of the norms of practical reason, which is a controversial assumption in this arena (see Wedgwood 2002). At the very least, this is not a mere extension of the teleological considerations of the sort found in Smith.

The second problem with holding instrumentalism is that Lenman’s brief considerations in favor of it aren’t compelling. He seems to fail to distinguish between a transition from a normative belief to the corresponding desire being governed by a norm versus the agent representing that norm in the content of a desire. Perhaps it is a conceptual truth that reasoning requires being governed by some norms. But it is not at all obvious or a conceptual truth that we must represent those norms in the contents of our desires in order for a psychological process to count as reasoning at all. Compare the analogous claim with respect to theoretical reasoning: one must believe, for example, that modus ponens is a good form of inference in order to be counted as employing it in any instance—a belief I presume many genuine reasoners lack. Surprisingly, at this point the Humean appears to be the one at risk of hyper-intellectualization. Apart from an independent argument for Lenman’s theory of practical reasoning, we have no reason to believe the extended teleological argument establishes motivational Humeanism.

Furthermore, even if we grant instrumentalism about practical reason as an acceptable assumption, this leads to a second problem with Lenman’s position. The truth of instrumentalism alone doesn’t necessarily entail that anything flouting the instrumentalist norm fails to count as rationalizing. As we have seen, a rationalizing explanation merely needs to show that there was something, from the agent’s perspective, to be said in favor of an action or attitude. As we’ve seen, a normative belief (e.g. that there is reason to help these workers) meets this condition. Where the issue is motivation, Humeans and rationalists are meant to be neutral on what the norms of practical reason are. Even motivational rationalists could assert instrumentalism about practical reason. Such a theorist could simply hold that people like Roberta can act irrationally, since they fail to satisfy any antecedent desires, though their actions are caused and explained in a way that makes sense or is intelligible (unlike the parsley or radio cases).

Such teleological considerations by themselves remain unable to motivate universal attribution of antecedent desires. However, Humeans might appeal to teleology in a different manner, as Stephen Finlay (2007) does in his Argument from Voluntary Response. Like Lenman, he doesn’t strictly speaking deny that normative beliefs can cause desires; he rejects the idea that “such causation is ever an instance of motivation” in his special sense (p. 224). The idea is that mere causation is not sufficient for genuine motivation—failing to amount to an instance of agency or teleology, or a voluntary “response to normativity.” And the Humean flavor of the view comes from the positive account of voluntary responses: “the formation of desire (adoption of a new end) can only be voluntary if it is motivated by some further desire” (p. 233).²²

While Finlay raises a number of worries for his opponents, his main qualm is that they paint rational motivation as too passive. By all accounts, we should not develop a conception of rational motivation as merely causal, putting it on par with, say, reflexes. But if Roberta simply desires to help the workers after believing she ought to do so, this seems to be nothing but mere causation. Voluntary actions, Finlay insists, “all proceed from desiring some end,” yet “the causal processes by which desires themselves are initiated [rather than actively formed] are merely causal and not teleological” (p. 234). But what more do we need to capture this teleology? In Roberta’s case, Finaly believes it’s implausible that she would want to help the reform efforts without a further desire playing a role—namely: “desiring that the workers not be exploited” (p. 234). Reflecting on the issues and considering such cases, we might then be lead to an account of voluntariness requiring antecedent desires: “In order for a new end to be generated voluntarily from the perception of a reason, it must be motivated by a desire that is not itself generated from that perception” (p. 234).

But I submit that a Humean view of voluntariness like Finlay’s is question-begging, or dialectically unhelpful (to use a phrase that is hopefully less vexed). There are certainly various plausible things to read into a case like Roberta’s. But there are many different cases of this kind, and they cannot all be explained by a desire for workers to not be exploited. We need a more general kind of reason for positing the relevant antecedent desires. Finlay provides this, but the support for it is wanting. As we’ve seen, there is nothing puzzling about explaining someone’s action ultimately in terms of their belief that it’s the right thing to do. Roberta helped voluntarily because she wanted to help, and this counts as a response to normativity since she ultimately did it because she thought it was the right thing to do. Adding another desire seems entirely unnecessary, pressing only to those who are already convinced of the Humean theory of the structure of motivation. This may seem to merely result in a dialectical stalemate, but the presumptive case for rationalism (from Section 2) is meant to avoid this. Given the ubiquity of ordinary explanations like Crist’s, which don’t appeal to antecedent desires, there are independent grounds for breaking such a stalemate in favor of rationalism.

Once again, teleological considerations alone can only establish that voluntarily A-ing requires a preceding desire to A. The extended teleological arguments don’t ultimately provide independently compelling reasons to go further and posit desires at the beginning of all motivational chains.

3.3 Practical Reasoning

Similar to the appeal to the teleological nature of motivation, Humeans might point to the structure of practical reasoning to support their account. After all, figuring out what to do quite often involves determining the means to our antecedent ends; perhaps that’s all it involves. Simon Blackburn, a devout Humean, takes this tack explicitly in his paper “Practical Tortoise Raising” (1995/2010), which commemorates Lewis Carroll’s “What the Tortoise Said to Achilles” (1895) on its centenary.²³ Blackburn says the question he wants to ask is “whether the will is under the control of fact and reason, combined,” which he says is an issue of “cognitive control, or control by the apprehension of fact and reason” (p. 7). His “Humean conclusion,” however, is that “there is always something else, something that is not [ultimately] under the control of fact and reason” (p. 7). Blackburn (as the Humean tortoise) considers and rejects several arguments (from the Kantian Achilles) for the allegedly opposing claim: that “decision making is [ultimately]… under the control of [one’s apprehension of] fact and reason” (p. 21). Of course, our Kantian or rationalist doesn’t hold to this strong of a claim if it is read as a universal generalization. Motivational rationalists only hold that sometimes actions are ultimately under “cognitive control” or the “apprehension” of certain normative propositions. However, Blackburn seems to take his opponents to be those who hold this weaker claim as well.²⁴

The structure of Blackburn’s argument is largely defensive. He considers four main opposing arguments. The first (which is similar to the primary style of argument present in Carroll’s Achilles) attempts to add more beliefs to the relevant argument; the second appeals to decision theory; the third invokes general Kantian considerations of impartiality; and the fourth appeals to the instrumental principle. Blackburn’s defense against all but the first strategy seems to involve legitimate points (though I doubt any rationalist would advance them anyway). But his discussion of the first strategy is instructive, especially since it attaches to a common thought implicit in many proponents of Humeanism. Blackburn attempts to show that adding further beliefs will not force the tortoise to move from the basic premises of something like the practical syllogism to the conclusion (the intention or action). The initial “argument,” according to Blackburn (p. 8), is:

• 1. I would prefer eating lettuce to eating souvlaki.

• 2. The moment of decision is at hand.

• 3. Let me choose to eat lettuce rather than souvlaki!

His starting idea is that we may accept these premises as true (and even the conclusion if we think of it as the proposition that I should or will make the choice), but this does not guarantee that I will make the choice by forming the corresponding intention. What Blackburn thinks we need is the desire to do so. In other words, believing that I have the desire (i.e. 1) and believing that the opportunity to satisfy it has arrived (i.e. 2) will not guarantee that I have the desire for lettuce rather than souvlaki (or perhaps that I have the desire to choose to eat this bit of lettuce in front of me). Even if we add the normative belief (“I think it is right to prefer lettuce to souvlaki” p. 8), we still aren’t guaranteed that my “will is determined” accordingly. The same goes for the belief that lacking the relevant preference is irrational, and so forth.

Now Blackburn’s discussion is difficult to fit precisely into our dialectic. But there seems to be the materials for a kind of argument that is at least implicit in his discussion and in many Humean conceptions. It starts with something like the following idea (I purposely leave it rather vague), which is typically based on considerations having to do with the teleological nature of desires (a la Smith 1987):

Prior Desire: desires are the only goal-directed states, so one can’t get motivated by anything other than such a state.

This thought can then be applied to any case of a belief’s seeming to produce, in a rationalizing way, an ultimate desire. We simply apply Prior Desire to yield the conclusion that there must have been some prior desire, such as the desire to desire what one thinks one should do. While this is distinct from the common ground that a subsequent desire must be present, it resembles Lenman’s (1996) generalization of Smith’s teleological argument from the previous section.²⁵

There are problems with this kind of argument for motivational Humeanism, however. One is that it can lead to an infinite regress of desires. (This is perhaps ironic in the context of Blackburn’s attempt to use this tortoise strategy to bolster a case for Humeanism.) On a certain unrestricted reading, we can always apply Prior Desire to any alleged final desire. For example, the Humean attempts to say that the virtuous person is motivated to do what she thinks is right only because she has a prior desire to do so. But the same considerations apply to that more basic desire. How did one get it? That is, how did the agent come to be motivated to do what she thinks is right? On a certain reading of Prior Desire, we must tack on another prior desire. But then we are off on an infinite regress (and one that looks vicious).

Of course, Humeans will emphatically reject any such reading of Prior Desire. They will attempt to clarify it to allow for the possibility that desires can simply come into existence independent of prior desires and play the relevant rationalizing role. But, no matter how one clarifies Prior Desire, the point so far is that Humeans will have to do it in a way that allows for this possibility. That is, they must hold that some desires can simply come into existence and rationalize without the aid of prior desires while beliefs can’t. But once that is granted, there is no principled reason to deny it to the derived desire with which we started that the rationalist says is produced and rationalized by a belief. Why not say that the belief just produced the desire without any need for some prior one?²⁶

Certain initial characterizations of Prior Desire (and to the elusive argument suggested by Blackburn’s tortoise-style considerations) might have some intuitive plausibility. But, when scrutinized, unless it is a question-begging statement of Humeanism, it amounts to nothing more than an ad hoc account. It could of course still be true, but no such tortoise-style considerations are dialectically helpful enough to provide a substantive case for Humeanism. The moral we should draw from the sort of considerations that Lewis Carroll originally raised are that the regress cannot be stopped even by positing some special psychological state. The solution in both the theoretical and practical case is to deny the assumption that a solution to the problem must take this form. As Nagel says, “the temptation to postulate a desire at the root of every motivation is similar to the temptation to postulate a belief behind every inference” (1970/1978, p. 30).²⁷

4 Conclusion

Despite recent criticisms, there is little reason to doubt that a rational agent’s normative beliefs can generate desires to act in accordance with them without serving or furthering antecedent desires. This is entirely compatible with the teleological nature of motivation and the structure of practical reasoning. Moreover, we needn’t cut the rationalist’s view out of our theorizing in the name of parsimony.

Once we recognize that our beliefs about what we ought to do frequently play a role in motivation, Humeans face the problem of providing revisionary accounts of this role. We have every reason to think, for example, that we sometimes explain and rationalize actions by ultimately citing only a normative belief. The burden of proof is on Humeans to show why we should consider such explanations erroneous or disguised attributions of an antecedent desire. Given the failure of the key theoretical arguments, this is a heavy burden we may expect they cannot meet. Similar to the situation of psychological egoism, this provides at least a presumptive case in favor of the alternative view: motivational rationalism. Perhaps, then, we can in fact be ultimately motivated to act for the pure and simple reason that we believe it’s the right thing to do.²⁸