Realism, Rational Action, and the Humean Theory of Motivation

Abstract

Realists about practical reasons agree that judgments regarding reasons are beliefs. They disagree, however, over the question of how such beliefs motivate rational action. Some adopt a Humean conception of motivation, according to which beliefs about reasons must combine with independently existing desires in order to motivate rational action; others adopt an anti-Humean view, according to which beliefs can motivate rational action in their own right, either directly or by giving rise to a new desire that in turn motivates the action. I argue that the realist who adopts a Humean model for explaining rational action will have a difficult time giving a plausible account of the role that desire plays in this explanation. I explore four interpretations of this role and argue that none allows a Humean theory to explain rational action as convincingly as an anti-Humean theory does. The first two models, in different ways, make acting on a reason impossible. The third allows this possibility, but only by positing a reason-sensitive desire that itself demands an explanation. The fourth avoids this explanatory challenge only by retreating to an empty form of the Humean view. In contrast, an anti-Humean theory can provide an intuitively plausible explanation of rational action. I conclude that the realist about reasons should adopt an anti-Humean theory to explain rational action.

According to the realist, reasons for action are truths about which considerations count in favor of acting. While realism has the benefit of capturing certain common intuitions about the objectivity of reason judgments, it faces the challenge of explaining how these judgments motivate rational action. For the realist, judgments about reasons are beliefs. The worry is that there is a gap between beliefs about reasons and rational action that realism has difficulty crossing in an intuitively plausible way.

If the realist is to explain how such beliefs give rise to rational action, she needs to supplement her account of normative reasons with an account of motivation that specifies which psychological elements are responsible for generating rational action.¹

One familiar option is the Humean theory of motivation, according to which belief in a reason combines with an independently existing desire to generate rational action. An anti-Humean theory, in contrast, claims that belief in a reason can generate rational action in its own right, either directly or by giving rise to a new motive.

It is often assumed that the Humean theory allows the realist to explain rational action as well as the anti-Humean theory does. Indeed, the Humean theory is thought to have the explanatory advantage since it appeals only to desires and ordinary, non-motivating beliefs. In what follows, however, I shall argue that the Humean theory has a surprisingly difficult time providing a plausible account of the role that desire plays in the explanation of rational action. I’ll explore four interpretations of this role and argue that none allows the Humean theory to explain rational action as naturally and plausibly as the anti-Humean theory does. I shall not be concerned to defend a realist conception of practical reasons here; instead, I shall focus exclusively on the question of how a realist might best explain rational action.

1 Rational Action

When acting rationally, an agent undertakes to act in light of her belief about what she has reason to do. She chooses her action because it is supported by reasons. In this sense, rational action seems to embody a distinctly rational form of motivation in which the agent guides herself by the thought that an action is recommended by reason. This guiding thought need not always be explicitly articulated. For rational action to be possible, however, the agent must, at some level of awareness, conceptualize the features to which she is responding as reason-giving. This is necessary to distinguish cases in which the agent simply happens to respond to reason-giving considerations, by accident or without recognizing them as such, and cases in which she responds because she recognizes that they are reason-giving.

To respond directly to a situation’s reason-giving features, but without responding to them as reason-giving, would be to miss the fact that they rationally merit one’s response. Consider, for example, an agent who comes to recognize that quitting smoking would increase his lifespan. He might either respond to this fact simply because he is motivated to do so, or he might respond because he recognizes that this fact is reason-giving. Only in the latter case, I think, would his action count as distinctly rational in its motivation. If this is right, then a guiding thought regarding a consideration’s status as reason-giving is necessary for rational action. The task for the realist view of practical reasons is to provide an intuitively plausible explanation of how acting on one’s recognition of a reason in this sense is possible.

2 Two Rivals Explanations of Rational Action

I’ll begin with a brief outline of the central differences between a realist account of reasons that explains rational action using a Humean theory of motivation and one that explains rational action using an anti-Humean theory of motivation. Prichard and Ross are realists who appear to accept the first theory of motivation, Nagel a realist who accepts some version of the second.²

Take the Humean theory first, which many consider to be the orthodox account of both the nature of propositional attitudes and the explanation of intentional action. This view appeals to a version of the belief-desire model to explain rational action. I think that there may actually be two different versions of the Humean theory of motivation in the contemporary literature, the more traditional version, which closely resembles Hume’s own account of the respective roles of belief and desire in the explanation of intentional action, and a version advocated by Michael Smith ³, which I characterize very briefly below. I shall focus on the more traditional version here since I think that it is the view that philosophers typically have in mind when discussing the Humean theory of motivation (I think this is true of Prichard and Ross, for example).⁴

In Book II of the Treatise, Hume claims that reason and belief are motivationally inert, and that in order to generate intentional action they require the help of an independently existing motivational state (i.e. desire) .⁵ On this account, belief and desire are distinct kinds of mental states with very different natures, belief being essentially representational, desire essentially motivational. Any intentional action must be the product of the causal combination of one of each type of mental state. Hume’s view has three core commitments, then: (1) belief cannot motivate, either by itself serving as a motive or by giving rise to a new motive,⁶ (2) belief and desire are modally distinct states, with different directions of fit, and (3) any adequate explanation of intentional action requires reference to both a means-end belief and a desire. (In contrast, Smith’s view includes commitments (2) and (3) but appears to allow that a belief with the right content can generate a new motive.)⁷

In order for a realist to make use of this theory of motivation to explain rational action, she needs to add a distinctive sort of belief and a matching desire to the picture. Hume, of course, denies that there are truths about justifying reasons that can be represented in belief. Instead, he appears to think that human action is caused by a combination of standing desires and non-normative theoretical beliefs about means-end connections. The first item the realist needs to add to the Humean theory of motivation, then, is a belief that represents truths about reasons, i.e. a normative belief. In addition, she’ll need to find a desire with the right content to enable this belief to motivate rational action.

What is the right content for the desire? While a normative belief might combine causally with any number of desires to produce action, only a desire whose object is to do what reason requires or recommends will allow acting for the sake of a reason. To see this, consider the normative belief that reason requires maximizing the general welfare and a desire whose objects happens to be maximizing general welfare (but not because reason says to do so). Despite the similar content of the belief and desire here, the agent who acts on this belief and desire will not act because reason requires it. She will in fact do what reason requires (i.e. maximize welfare) but not because reason requires it. Her action will be in mere external conformity with reason. The desire necessary for rational action, then, seems to be one with the content of doing what reasons recommends or requires.⁸ This is the analogue, for the case of rational action, of the desire that Prichard and Ross appeal to in the moral case, i.e. a desire to do one’s duty.⁹ With this desire identified, the explanation of rational action runs as follows: among the rational person’s standing desires is a desire whose object is to do what reason recommends, and distinctively rational action results when this desire is triggered by a normative belief in a justifying reason. Interestingly, while the desire needed to explain rational action must have doing what reason recommends as its object, on the Humean account both the existence and the content of this desire must be explained by something other than belief’s direct causal influence. It must exist apart from this influence if it is to provide an independent source of motivation with which the belief can combine to produce rational action. And while its content must match that of the normative belief, the fact that it does so cannot be the direct result of this belief. I shall return to these points below.

With a normative belief and a matching desire added to the standard Humean belief-desire model, then, the realist can use the Humean theory of motivation to explain rational action. On this account, beliefs about reasons for action are themselves motivationally inert but can combine with an independent, pre-existing desire with the right content to generate rational action.

According to the anti-Humean theory of motivation, in contrast, the recognition of a justifying reason can generate rational action in its own right, either by motivating action directly or by producing a new desire that motivates rational action. A rational agent, in turn, is one who is responsive to belief’s motivating power in the sense that there is nothing in her that blocks this motivational force. It is worth noting that the anti-Humean need not claim that this belief necessarily motivates, since outside interference can block its ability to motivate. The claim is simply that the recognition of a reason can give rise to new motivation in its own right, without the help of a pre-existing desire.

With this initial sketch of the two rival accounts of how the recognition of a justifying reason can motivate rational action, it may seem that the Humean account of motivation has the advantage on explanatory grounds. It allows a close, if contingent, connection between beliefs about justifying reasons and motivation; and it is consistent with familiar intuitions from the philosophy of mind and action. In contrast, the anti-Humean account may seem to grant beliefs unusual motivational powers. Why, then, shouldn’t a realist about justifying reasons accept a Humean theory of motivation?

In laying out the contours of the Humean explanation of rational action and identifying the content of the desire involved, I have taken for granted that it is obvious what sort of explanatory role the desire plays in generating rational action. In fact, however, I think that it is quite unclear what this explanatory role is supposed to be. Since in the Humean account desire’s explanatory role is more often taken for granted than articulated clearly, my argument will proceed by exploring four possible models for understanding this role, each suggested by weaknesses in the preceding model. I’ll argue that each suffers from a distinctive kind of explanatory failure. The first two fail because, in different ways, they make acting on a reason impossible. The third may be able to overcome this worry, but only by positing a reason-sensitive desire that itself calls out for explanation. The fourth signals a retreat to an empty form of the Humean view.

3 The First Interpretation of Desire’s Role: Desire as a Motivational/Justificational Disposition

One way to understand desire’s explanatory role within the Humean theory is to think that it provides rational motivation with a roughly instrumental character. On this account, the desire provides not only the motivational impetus towards action but also in some sense the rationale for action. As I’ll argue, however, in the end this model gives desire not only an explanatory but also a justificatory role and so should be rejected as an interpretation of the Humean theory of motivation. However, because it may come to mind as the form that a belief-desire explanation might take, and because it provides a useful contrast to the other three interpretations, it will be helpful to explore it before moving on.

The thought behind this first interpretation is that in motivating rational action a desire not only moves the person to act but in a deeper sense explains why he acts. On this way of understanding desire’s role, all action has a broadly instrumental character in the following sense. While the object of the desire determines the result to be brought about, the desiring provides the motive and, through this, the rationale for acting. In this account, the object of a desire is not the ultimate motivator and justifier of action. If not for the fact of being desired, the object would be both motivationally and justificationally irrelevant to action.

It should be clear why this conception of desire’s role is an unacceptable account of motivation for the realist. In order for acting on a reason to be possible, the object of the desire to do what reason recommends must provide the ‘why’ or ‘rationale’ of rational action. Only then can a rational agent act directly on her justifying reasons when she acts from her normative belief and her desire to do what reason recommends. On the current interpretation, however, this doesn’t seem possible since desire’s role seems to create motivational and justificational interference with the content of the normative belief, preventing it from determining action. Action so conceived is always ultimately desire-based, not reason-based, and so acting on one’s recognition of a reason isn’t possible.

This worry might seem misplaced since the intentional object of the desire that motivates distinctively rational action on this account is doing what reason recommends. However, the problem is not with the content of the desire but with the way that the desire serves as a motivational ground for action. Even in the case involving a normative belief in a reason and a matching desire, on this interpretation the agent acts because he desires to do as reason recommends, not because reason recommends it. The object of his desire – doing what reason recommends – is never able to motivate action directly. The result is that the agent can never act on a justifying reason simply because it is a justifying reason. The rationale with the distinctively instrumental shape gets in the way and changes the ground of the action.

The problem with adopting this conception of the role of desire in the explanation of rational action, then, is that it is not innocent of normative commitments. On this account, what happens when reason-based motivation is assimilated to belief-desire motivation is that we inadvertently end up with the addition of a different kind of reason, a desire-based reason for action, which prevents the original justifying reason for action from playing its proper role in the generation of rational action. The first interpretation of the Humean theory of motivation must be rejected, then, for it seems to incorporate elements of a Humean view of normative reasons into its theory of motivation. When paired with a realist theory of normative reasons it has the consequence that one can never act directly on the justifying reasons represented in one’s normative beliefs.

4 The Second Interpretation of Desire’s Explanatory Role: Desire as a Reliable Causal Link

The role of desire in the Humean explanation of rational action needs to be purely explanatory, not justificatory, then. Desire must provide a bridge from the normative belief to action in a way that does not introduce any justificatory interference. The dispositional account of desire seem to meet this requirement easily, for it conceives of desire simply as a disposition to respond motivationally to the content of a belief. On this view, desire is a function that takes a normative belief as the input and reliably delivers motivation as the output.

With this purely motivational, nonjustificatory role for desire, the explanation of rational action runs as follows. A belief that an action is recommended by reason will, in someone who desires to do what reason recommends, give rise directly to a motive to do the action. Motivation on this account is just the result of the triggering of a disposition to be moved to action by a belief in a justifying reason. Thus this conception of motivation seems to allow explanations in which the agent can act directly on her justifying reasons without motivational or justificational interference.

On this model, desire is a reliable motivational link between normative belief and action. Notice, however, that as so far described the account places no conditions on how, or in virtue of what, the reliable link is forged; the only requirement is that it be reliable. This raises a worry, for there are any number of potentially reliable motivational connections between recognized reasons and action, many of which do not involve the distinctly rational motivation necessary for genuinely rational action.

To see this more clearly, consider two examples in which a disposition might provide a reliable motivational link between normative belief and action, but a link of the wrong sort. Both involve the disposition’s reliably producing action in virtue of being sensitive to the wrong aspect of reason judgments, that is, to something other than their content. This might happen if the desire were sensitive to some extraneous feature of reasons judgments, for instance, the fact that all reason judgments in a certain language begin with the same combination of consonants. Then normative beliefs might reliably trigger an independently existing disposition and so produce action, but this disposition would be sensitive to the wrong aspect of these beliefs. Or consider a reliable disposition to respond that is the product of hypnosis. In this case, too, while belief in a reason would reliably trigger a motivational response, the disposition would not be appropriately responsive to the justificational authority of the reason, but instead simply a reflection of the psychological effect of hypnosis. The specific content of the normative beliefs would make no difference to the motivation. This disposition, then, like the one above, would fail to explain acting on a reason in the right way. While it would yield external conformity with a reason judgment, or perhaps the wrong sort of internal conformity, it would not motivate action for the sake of the reason recognized.

The problem in both examples is that the disposition involved could reliably cause behavior in conformity with recognized reasons but not produce genuinely rational action. This is a consequence of the fact that the disposition is conceived of as merely a reliable motivational link. The possibility of genuinely rational action, in contrast, requires that action be motivated in the right way, by a recognition of the normative authority of the justifying reasons, not by extraneous features of the belief.

The upshot seems to be that while the Humean theory of motivation needs to avoid giving desire a justificatory role that might interfere with the recognized justifying reasons, it cannot conceive of the desire’s role simply as providing the means for a reliable motivational connection between representations of justifying reasons and action. The desire must also be able to track representations of justifying reasons in virtue of their rational authority and motivate in virtue of this.

5 The Third Interpretation of Desire’s Explanatory Role: Desire as a Distinctively Reason-Sensitive Causal Link

As the discussion above makes clear, what’s needed is a disposition that responds to reasons as reasons. This sort of motivation would be transparent to the content of the normative beliefs, allowing justifying reasons to serve as the person’s motivating reasons for action. With a disposition so conceived, it seems that recognized reasons might figure in the right way in generating rational action. Unlike in the first interpretation, which gives desire a competing justificatory role, desire here plays a purely motivational role; but unlike the second interpretation, which fails to give the desire the right kind of reason-sensitivity, this interpretation makes the desire sensitive to the right aspect of the reasons, i.e. their justificational authority. So the explanatory role attributed to desire within this third interpretation seems to be of the right sort to generate genuinely rational action.

While this account of desire’s explanatory role is clearly an improvement over the first two models, it suffers from weaknesses of its own. I’ll discuss three. To begin, it is not clear that the Humean account really has the resources to distinguish a blind reliable motivational link from a genuinely reason-sensitive one if desire is defined by its functional role, which is a bare motivational role. The most natural strategy for drawing this distinction would be to attempt to build something more into the content of the desire that would allow it to track reasons in the right way, for example, a content of doing what reason requires because reason requires it. It is difficult to see how a desire could have this content, however – a content regarding how it should track reasons – or have it in a way that affects how the desire actually responds to reasons. Perhaps one might instead try to explain the desire’s reason-sensitivity by delineating its counterfactual relations with the normative belief. It’s not clear, however, which counterfactual dependency relations could establish the right relationship to render the desire a desire to act on reasons because they are reasons. Here again it seems that the most one could get would be a tight correlation. Since I may be suffering from a failure of imagination regarding this point, however, I do not want to rest my case against the third model on it. Instead, I want to grant for the sake of argument that the Humean can overcome this problem and turn to two remaining worries about the third model.

The reason-sensitive desire is supposed to play its explanatory role by serving as the original source of rational motivation. However, once the content of the desire simply mirrors the content of normative beliefs, as it does on this interpretation, it is no longer clear that the desire plays an independent explanatory role. On the reason-sensitive account, (1) desire is merely a disposition to respond motivationally to recognized reasons, (2) it plays this role in virtue of being triggered by a normative belief, and (3) it need have no phenomenological component. Given these features, I would argue that there is little reason left to claim that the desire is an independent existence. Such a thinned down dispositional state, whose only job is to be triggered by representations of reasons, is more plausibly described as a normative belief’s causing action directly. Once we arrive at a form of deeply reason-sensitive motivation that is simply the reflection of normative beliefs, an anti-Humean account should be favored on grounds of parsimony.

Finally, even if we grant both that the desire can provide the right sort of reason-sensitive reliable link between normative belief and action and that desire so understood can serve in a non-trivial sense as an original source of motivation apart from reason’s direct causal influence, the Humean theory faces the challenge of explaining both the existence and the deep reason-sensitivity of this independent disposition. Oddly, while this desire doesn’t explain much, its presence in the account gives rise to a new explanatory burden. What needs explaining is the origin of this distinctively reason-sensitive kind of motivation. According to the Humean this reason-sensitive motivation is not itself the direct motivational product of the causal power of reason or normative belief; instead, it is a motivational tendency that is there anyway in the human psyche, ready and waiting to take up the job of enabling otherwise inert representations of reasons to motivate action. The worry is that any account of this reason-sensitive motivation that doesn’t credit reason¹⁰ as its cause is going to be much less plausible than an account that does.¹¹

To better appreciate the challenge, consider the nature of the motivational responsiveness that needs to be explained. In order to provide even the basic features that we expect to find in reason-responsive motivation, the desire that explains rational action needs to be an extremely flexible and intelligent motivational tendency. Part of responding appropriately to justifying reasons as reasons seems to involve not simply registering the fact that they are ‘reasons,’ under a generic label, but responding to them in their particular varieties and strengths. Given that reasons differ in stringency and kind, motivational responses should too. Thus, it seems that the motivation generated by an enticing reason to value something is importantly different from the motivation generated by a reason that requires or forbids an action. Further, in many cases rational motivation should reflect the subtleties of a person’s understanding of why considerations are reasons. Recognizing a duty to help another presumably involves not simply registering the bare thought that helping is required by reason, but instead a more nuanced understanding of the reason why helping is a duty. This understanding should be reflected in motivation, so that motivation changes in response to subtle changes in reasons for acting. In short, the motivational responsiveness involved in rational action is not a simple, uniform response to the representation of a reason. It is a multi-faceted, fine-grained sensitivity to justifying reasons in their variety and complexity.

The anti-Humean theory of motivation has a natural way to explain these features of reason-sensitive motivation. On this account, since motivation is both normatively and causally under reason’s control, such complex motivational responsiveness is to be expected. We respond in a nuanced, flexible, and complex fashion because that’s the way reasons are, and our recognition of them causes our motivations, barring interference. In contrast, the Humean theorist of motivation has to find some source other than judgments about reasons for these features of motivation. The obvious contenders are not, alone or together, nearly as plausible as the anti-Humean’s straightforward explanation that recognized reasons directly give rise to our motivational responses. The perplexing question is how a disposition that is not itself caused by reason would have this complexity, flexibility, and nuance. It is difficult enough to imagine how an independent disposition could track just the bare information that something is a reason, let alone all of these other features of reasons, which we routinely expect to see manifested in rational motivation.

The two obvious sources for the Humean to appeal to in order to explain the disposition involved in rational action are education and nature. Education could provide a way to exert at least indirect control over the motivational system in light of what reason tells us about what our justifying reasons are. However, since on the Humean view the disposition to respond to reasons is not subject to the direct causal control of reason, education would have to be conceived of entirely as a matter of a nonrational habituation of the motivational structure. While some education of our motivational tendencies may take this form, it is implausible that the broad ground-level sensitivity to recognized reasons would have its source here, in part because the very project of education seems to presuppose it. Further, our motivational responsiveness to reason is pervasive, and present apart from education. It’s not as if we require early, careful training to be able to act on reasons, as this account would seem to require. Even those with very little education have no difficulty negotiating the basic terrain of acting on reasons. Education cannot be the sole or even the primary source of reason-responsiveness, then.

A more plausible suggestion is that education builds upon a natural disposition to respond motivationally to recognized reasons. The first problem with this proposal is that it is difficult to get a clear sense of what this would involve since a motivational tendency to respond directly to recognized reasons that is not itself caused by reason looks odd from the start. Everything else in nature appears to be causally responsive in a non-reason-sensitive manner. Human motivation, then, would have to be unique in being naturally sensitive to representations of justifying reasons but sensitive independently of the direct causal influence of reason. Even if this idea could be made more intuitively plausible, it requires what seems to be an unnecessary extra step in the explanation, positing an independent desire to respond to reasons when the simpler and more plausible explanation would be that reason is the kind of faculty that can both recognize reasons and govern action directly as a result of this.

This argument can be strengthened. Following Nagel, we can distinguish between what appear to be two different kinds of desires: those that exist apart from reflection and those that exist only as a result of reflection (what he calls unmotivated and motivated desires, respectively).¹² Paradigmatic examples of desires that precede reflection are ones with a distinctive phenomenology that are tied to physiological conditions, such as the desire for food. Since these desires are related to physiological conditions that exist independently of reason, it at least makes sense to think that they too exist independently of reason. However, it does not in the same way make sense to think that reason-sensitive motivation exists independently of reason’s direct causal influence. It is only by mistakenly taking physiologically-generated desires as the paradigm, I think, that it could seem to make sense to think that reason-sensitive motivation exists independently from reason and is merely triggered by the information reason provides.

Notice, further, that when we reflect on desires generated by bodily conditions and judge that they provide us with reasons, the resulting motivation seems to be of a distinctively reflective sort that is quite different from that involved in the original, physiologically-generated desire. This reflection-following motivation is most plausibly viewed as reason-generated.

We can distinguish, then, between two types of motivational states: those that are generated by physiological conditions and that exist independently from the activity of reflection, and those that are reason-generated. It this is right, it is a mistake to regard all motivation as having a source outside of reason, such that even deeply reason-sensitive motivation must be explained by something other than reason’s direct causal influence. Only desires associated with definite physical needs are plausibly thought to exist apart from reason; reason-following desires, in contrast, are more appropriately viewed as produced by reason itself. Indeed, if reason didn’t produce the latter, it seems that nothing could. The Humean view gains an unearned air of plausibility by assimilating all motivation, even clearly reason-following motivation, to the model of physiologically-generated desires. For the reasons I have cited, it is more plausible to think that there is a basic distinction between two quite different kinds of motivational states.

At this point the Humean theorist might be inclined to try to counter-balance the difficulty the theory has in explaining the nature and existence of the reason-sensitive motivation by citing what Humeans often regard as a unique explanatory strength of their view, i.e. that it accounts well for commonly observed failures of reason judgments to motivate. But of course the anti-Humean theory has its own account of motivational failures, i.e. that they result from various sorts of interference with reason’s ability to motivate. The existence of motivational failures does not, then, automatically count as evidence for the Humean view and against the anti-Humean view. Further, even if the Humean view did have a very natural account of such gaps, this would not diminish the fact that the Humean account provides a much less plausible explanation of the basic positive link between recognized justifying reasons and motivation.

To sum up, then, the third model of desire’s explanatory role suffers from three problems. First, in order to avoid the bare reliable link conception of the disposition to do what reason recommends the Humean theory embraces a conception according to which the disposition is genuinely responsive to reasons. It is not clear, however, that a view that conceives of desire in purely functional terms has the resources to distinguish a blind disposition from a genuinely reason-sensitive disposition. Second, even if the view is able to explain how the disposition is genuinely reason-responsive, it is not clear that the disposition so conceived is doing real explanatory work. Third, even if the first two worries can be met, the third model faces the challenge of explaining the origin and shape of such a disposition apart from the direct influence of reason. Not only is there good antecedent reason for skepticism about success here given the nature of the undertaking, but the rival anti-Humean hypothesis seems much more plausible.

6 The Fourth Interpretation of Desire’s Explanatory Role: Desire as the Abstract Quality of Being Such That One is Motivationally Responsive to Recognized Reasons

One final possibility that might appear to be available to the Humean in the face of these worries is the following.¹³ He might claim that when he talks about a disposition to be motivated by representations of justifying reasons all that he means to indicate is the fact that the person in question has the feature of being such that she is motivationally responsive to representations of justifying reasons, where this is not meant to entail that the person is in a definite mental state the existence of which requires explanation. On this interpretation, ‘desire’ is used to indicate possession of the abstract quality of being such that one is motivationally responsive, nothing more. No account of the source or mechanism of the sensitivity is part of the view; indeed, this is what allows it to escape the explanatory worries that plague the third interpretation. Its central claim is simply that one must be such that one is motivationally responsive in order for one’s representations of justifying reasons to lead to rational action.

This form of retreat marks the failure of an already ailing account, for on this interpretation the view is clearly empty. Anyone who thinks that we can somehow be motivated in the face of our recognized reasons, even the staunchest anti-Humean, will attribute to rational agents the abstract quality of being such that they are motivationally responsive to reasons (in virtue of something). The dispute is over what makes this responsiveness possible, and labeling the fact of responsiveness does not address this issue. The original claim of the Humean theory was that desire, taken to be a real mental state with distinctive features, had both a necessary and distinctive role to play in the explanation of how the representation of a reason can motivate rational action. Retreating to the idea that there is a need for motivation in any explanation of rational action does not contribute anything of substance to the dispute. That no one can escape affirming this need, abstractly characterized, cannot count as a reason to accept the Humean theory.

I have argued that adopting the Humean theory of motivation would make the task of explaining rational action more difficult for the realist. I have tried to show this by exploring four ways to understand the explanatory role of the desire in the Humean account and arguing than none of these allows a plausible explanation of rational action. The first model, which assigns desire a motivational-justificational role, gives all motivation an instrumental shape and so makes acting on a recognized justifying reason impossible. The second model, which conceives of desire as a reliable motivational link, also makes rational action impossible because the desire posited does not respond to reasons as such. The third model, which conceives of the desire as a reason-sensitive reliable link, might allow genuinely rational action if it can find a way to distinguish between the right sort and the wrong sort of reliable link, but even then it faces an additional explanatory problem: it needs to provide a plausible explanation of the existence and nature of a deeply reason-sensitive form of motivation without appealing to the causal power of reason. The fourth model, which can be seen as an attempt to avoid this explanatory challenge, merely posits the abstract quality of being such that one is motivationally responsive to reasons, without appealing to a definite mental state. The problem with this model is that it is empty.

These considerations suggest that a realist may have a much easier time explaining rational action if she adopts an anti-Humean theory of motivation, assuming that the latter is not itself subject to damning objections.¹⁴ This is an interesting result, for while it’s often admitted that an anti-Humean theory of motivation allows the realist to respect important normative intuitions about the connection between reasons and motivation,¹⁵ it’s rarely thought that an anti-Humean theory has explanatory benefits over a Humean theory. If the line of argument in this paper is correct, however, then it has this advantage as well.¹⁶