1 The practicality of normative judgment

Suppose that you have indulged over the holiday period. You find that you have gained quite a bit of weight. You judge that you really ought to lose 10 lbs within the month to get yourself back in shape. You believe that you can only lose that 10 lbs in (what is left of) January if you go to the gym. Consequently, in light of both this belief and your normative judgment, you pick yourself up off the couch and head to the gym, unenthusiastic but resolute.

In this vignette, a normative judgment is cast in the role of a motivational state: your judgment that you ought to lose 10 lbs within the month combines with a certain means-end belief—your belief that you can only lose that weight if you go to the gym—to cause you to actually go to the gym. Just like a desire, your normative judgment combines with a belief to cause action.

Examples like this are often taken to be good evidence (Shafer-Landau 2003) for the practicality of normative judgment, the view that normative judgments are motivational states.Footnote 1 What it is for some mental state to be motivational, it seems, is for that state to be such that, taken together with some certain background belief, it can causally suffice for action.Footnote 2 And the above example is an instance of a normative judgment, taken together with a certain background belief, causally sufficing for action. Hence, it looks like normative judgments are motivational states. Certainly, in advance of our theoretical commitments pushing us this way or that, we ought to hold that normative judgments are motivational states. In other words, it should be agreed that the presumptive view here is that part of what it is for you to judge that you ought to F is for you to be motivated—to some degree, at least—to do F.Footnote 3 This is the practicality of normative judgment. Philosophers who endorse this doctrine (under other names or guises) include, amongst others, Mackie (1977), Darwall (1983), Gibbard (1990), Smith (1994), Blackburn (1998), Dreier (2015), and Copp (2018).

Normative realism is the doctrine that there are facts of the matter, independent of our minds, about what agents ought or have reason to do. The normative realist, however, has a problem accounting for the practicality of normative judgment. How so? Put simply, the normative realist is committed to normative judgments being beliefs—the doctrine known as ‘cognitivism about normative judgment’. And this is problematic because it is difficult to see how any belief could also be a motivational state. Indeed, according to the orthodoxy in philosophical psychology, beliefs and motivational states have different natures that cannot be reconciled together in one mental state (Anscombe 1957; Smith 1987, 1994; Parfit 2011).

Let us call this problem—that the normative realist faces in explaining the practicality of normative judgment—‘the practical problem’. It has its source in the fact that normative judgment, for the realist, must be both theoretical—that is, such that it aims at correctly representing normative reality—but also, on the face of it, practical—namely, such that it provides the judger with motivation for action. Viewed this way then, the practical problem is just the fact that, for the realist, normative judgments must be both theoretical and practical, but theoretical and practical modes of thought seem to have different natures that, on the face of it, cannot be realized together in one kind of thought. Believing that reality is some way or another, it seems, cannot by itself move you toward action. In short: the reality of the normative appears to require an impossible form of thought.

In this paper, I show how the normative realist can solve the practical problem—that is, coherently hold that normative judgments are both beliefs and motivational states, all consistent with the Humean theory of motivation. In other words, the realist can have her cake and eat it too: her normative judgments will be both beliefs and occupy the motivational role that such states prima facie seem to possess. The (low!) price to pay for this solution is the acceptance of a strong form of cognitivism about intention, the doctrine that your intentions to act are really beliefs about what you are going to do, and the making of a small—yet highly plausible—adjustment to our theory of what it is to be a motivational state.

The structure of the rest of this paper goes like this: in Sect. 2, I detail and motivate the practical problem. In Sect. 3, I introduce and motivate cognitivism about intention, the philosophical psychology upon which my solution to the practical problem rests. In Sect. 4, I introduce and explain my advertised solution. Lastly, in Sect. 5, I defend this solution from objections.

2 The practical problem

Normative realism faces a problem accounting for the practicality of normative judgment, the presumptive view that normative judgments are motivational states. Normative judgments, for the realist, must be beliefs. But beliefs, on the face of it, cannot be motivational states. Normative realism seems to require an impossible form of thought. This is the practical problem.

This way of conceiving matters is inspired by Smith’s (1994) formulation of the ‘moral problem’.Footnote 4 As Smith puts it: ‘…The idea of morality thus looks like it may well be incoherent, for what is required to make sense of a moral judgment is a strange sort of fact about the universe: a fact whose recognition necessarily impacts upon our desires (Mackie 1977). But the standard picture of human psychology tells us that there are no such facts. Nothing could be everything a moral judgment purports to be…’ (Smith 1994).

Let’s get clearer on the details of this problem. We have already seen why (we should presume that) normative judgments are motivational states: we often cite normative judgments in the explanation of action, explanations in which such judgments play the role of a motivational state. But why should we accept that normative realists are committed to normative judgments being beliefs? And, likewise, why should we accept that beliefs cannot be motivational states? I now address these matters in turn.

The normative realist is committed to normative judgments being beliefs because she is committed to the reality of normative facts—facts about what agents ought, or have reason, to do—that are independent of our minds. After all, if there are normative facts, then normative judgments must be the mental states that aim to correctly represent these facts. Beliefs are the variety of mental state that aim to correctly represent—or fit—the facts (Anscombe 1957; Smith 1987, 1994). Hence, for the normative realist, normative judgments must be beliefs. To judge that you ought to F, for the realist, is just to believe that you ought to F. This answers the first question.

Moving onto the second: why should we accept that beliefs cannot be motivational states? Simply put: because it is a consequence of the Humean theory of motivation. The Humean theory of motivation has it that action is only ever brought about by a belief and a desire working in tandem, that beliefs and desires are ‘distinct existences’ with no necessary connections between them, and that no belief is a desire (Smith 1987, 1994). Granting that what it is for some mental state to be motivational is for that state to be such that, taken together with some certain background belief, it can causally suffice for action, it straightforwardly follows from the Humean theory of motivation that beliefs are not motivational states. After all, if action can only be brought about by a belief and a desire working together, and if no belief necessitates the reality of any desire, and if no belief is itself a desire, it immediately follows that beliefs alone can never causally suffice for action. Since what it is for something to be a motivational state is for it to be able to causally suffice, in conjunction with some certain belief, for action, it follows that no belief can be a motivational state. In this way then, the Humean theory of motivation implies that beliefs cannot be motivational states.

But why should we accept Humean philosophical psychology? The Humean can justify her psychology by pointing out that it is really just part of our (empirically very successful) commonsense understanding of the mind. Suppose that you are outside and you realize that it’s going to rain. This belief, by itself, doesn’t move you, this way or that, in any way. You’re only prompted to, say, run inside if you don’t want to get wet. But if you desire more strongly to, say, jump in puddles, then you’re going to stay outside instead, because that’s where the puddles are. So, generalizing from this arbitrary example, commonsense psychology suggests that beliefs alone are insufficient to cause action and that action is only ever brought about by a belief and a desire working in tandem.

Our commonsense grasp of psychology also suggests that beliefs and desires are ‘distinct existences’ which bear no necessary connections to one another. After all, it seems that having one complete cognitive picture of the ways things are, rather than some other complete picture, doesn’t place any restrictions on how you are fundamentally motivated to act. Indeed, on the face of it, it seems like I could possess complete knowledge of the physical past, present, and future of the entire universe without being motivated to act in any way at all. Beliefs by themselves don’t cause, or necessitate, the reality of desires. Of course, my desire for ice-cream might combine with my belief that I can get ice-cream from Ben and Jerry’s to cause me to acquire the desire to go to Ben and Jerry’s. But this is consistent with beliefs alone having no causal or necessary connection to desires or to other elements of motivation. And this latter claim is all the Humean is affirming.

Lastly, our commonsense understanding of the mind suggests that no belief is also a desire. Beliefs and desires appear to have distinct, and irreconcilable, directions of fit (Anscombe 1957; Smith 1987, 1994). Beliefs, by their nature, aim to fit only the facts: your belief that your grandma is in Miami ought to be revised in light of evidence that she is actually in New York. Desires, in contrast, aim at making the world fit their contents, not at making their contents fit the world. So your desire that your grandma be in Miami ought not to be revised in light of evidence that she is actually in New York. Rather, in light of this evidence, this desire (if it is strong enough) ought to move you, in conjunction with certain means-end beliefs, to bring it about that your grandma is in Miami—for example, by prompting you to buy her a plane ticket from JFK to Miami International Airport. Commonsense suggests that beliefs and desires have different natures that, on the face of it, cannot be fused together in one mental state.

In sum, our commonsense grasp of the mental supports the Humean’s contention that action is only ever brought about by a belief and a desire working in tandem, that beliefs and desires are ‘distinct existences’ with no necessary connections between them, and that no belief is a desire. Hence, in advance of our theoretical commitments pushing us this way or that, we ought to endorse the Humean theory of motivation. It enjoys enough prima facie warrant to constitute the presumptive view in this region of philosophical psychology.

We are now in a position to understand the force of the practical problem. The normative realist faces a problem accounting for the practicality of normative judgment, the proposition that normative judgments are motivational states. She faces a significant challenge here because her normative metaphysics commits her to holding that normative judgments are beliefs, but the Humean orthodoxy in philosophical psychology looks to entail that beliefs cannot be motivational states. Since to claim both that normative judgments are motivational states and that they are not motivational states would be to fall into incoherence, the normative realist looks to be committed to denying either the practicality of normative judgment or the Humean theory of motivation. The bite of the practical problem lies in the fact that both of these doctrines are highly plausible, with their negations being correspondingly implausible.

We can also—following Smith (1994)—formulate the practical problem as a quartet of individually plausible but mutually inconsistent propositions (a ‘tetralemma’):

  1. (1)

    normative judgments are motivational states.

  2. (2)

    Only a belief and a desire working in tandem causally suffice for action.

  3. (3)

    No belief is a desire.

  4. (4)

    Normative judgments are beliefs.

(MOTIVATIONAL) mental state M is a motivational state =df M can combine with some certain belief to cause action

Proposition (1) expresses the practicality of normative judgment, propositions (2) and (3) articulate elements of the Humean theory of motivation, and proposition (4) states the realist’s commitment to normative judgments being beliefs (the doctrine known as ‘cognitivism about normative judgment’). Taken together with the definition of a motivational state in (MOTIVATIONAL), these propositions form an inconsistent quartet: if a normative judgment in conjunction with the right background belief can causally suffice for action, but only a belief and a desire can causally suffice for action and no belief is a desire, then not only must normative judgments not be beliefs, they must also be desires. But normative judgments, for the realist, must be beliefs.

The normative realist’s options here—given that it is a non-negotiable commitment of the realist that her normative judgments are beliefs—are limited, it seems, to denying either the practicality of normative judgment or some element of the Humean theory of motivation. Predictably enough, normative realists have typically responded to this dilemma in one of two ways: either by denying the practicality of normative or by denying the Humean theory of motivation. The first group of normative realists, those who deny the claim that normative judgments are motivational states, are known as ‘externalists’. For externalists, there is no internal necessary connection between normative judgment and motivation. The mere conjunction of a normative judgment and background beliefs can never causally suffice for action. Normative judgments are consequently not motivational states, by the lights of externalists. Merely judging that you ought to do F does not by itself give you any motive to do F. Rather, your normative judgment can only bring about action in conjunction with the right background desire—the desire, say, to do as you ought to do, or the desire to be kind, combined with the belief that you can be kind by doing what you ought to do, etc. Externalists include Railton (1986), Boyd (1988), Brink (1989), Parfit (2011), and Copp (2018).

The other main camp of normative realists, those who respond to the practical problem by denying the Humean theory of motivation, are known as ‘anti-Humeans’. They hold that normative judgments are both beliefs and motivational states. Consequently, on this alternative picture of the mind, beliefs alone can causally suffice for action: Tim’s normative judgment that he ought to pay his taxes can combine with his belief that he can pay his taxes by using TurboTax to move him to do just that. Anti-Humeans include in their number Nagel (1970), McDowell (1978, 1979), McNaughton (1988), Platts (1991), Wiggins (1991), Dancy (1993), Smith (1994), Scanlon (1998, 2013), Shafer-Landau (2003) and Wedgwood (2004, 2007).

There is a third option for the normative realist here: reconciliation. Some have claimed that we can reconcile normative realism with the practicality of normative judgment and the Humean theory of motivation. Indeed, Smith (1994) claimed that he could do that in the very same book in which he formulated the moral problem (that is, the inspiration for the practical problem as I present it here).Footnote 5 Given the plausibility of both the practicality of normative judgment and the Humean theory of motivation, a reconciliation is clearly to be preferred (ceteris paribus) over any picture in which one of these doctrines must be rejected.

An additional reason to favor a reconciliation here over a rejection of either the practicality of normative judgment or the Humean theory of motivation is that it effectively neutralizes one of the anti-realist’s principal arguments against normative realism. The anti-realist argues from truth of the practicality of normative judgment and the Humean theory of motivation to the falsity of cognitivism about normative judgment and realism itself (Blackburn 1998; Parfit 2011). Such an argument is clearly valid: if a normative judgment in conjunction with the right background belief can causally suffice for action (per the practicality of normative judgment), but only a belief and a desire can causally suffice for action and no belief is a desire (per the Humean theory of motivation), then not only must normative judgments not be beliefs (contra realism), they must also be desires (per non-cognitivist anti-realism). The anti-realist further maintains that the practicality of normative judgment and the Humean theory of motivation are firmer elements of the manifest image, deserving more of our credence, than the proposition that normative judgments are beliefs. Consequently, she concludes, we have decisive reason to hold onto the former two propositions and thus reject the latter. After all, the anti-realist proclaims, our normative judgments are more like a species of conative states, such as approval or disapproval etc., than an instance of belief. As Blackburn puts it: ‘…ethics [is] more a matter of knowing how (to behave), or knowing whom (to defer to, or punish, or admire) or knowing when (to act, or withdraw), than a matter of knowing that something is the case….’ (Blackburn 1998). Hence, the anti-realist concludes, we ought to prefer the practicality of normative judgment over cognitivism about such states. And likewise with respect to the Humean theory of motivation: it is deeply embedded in our commonsense understanding of the mind, even more so, she insists, than cognitivism about normative judgment. In this way then, the anti-realist can reformulate the practical problem as a powerful argument against normative realism and in favor of non-cognitivism. However, a successful reconciliation of the propositions making up the practical problem will immediately take the wind out of the sails of this anti-realist dialectic. After all, if cognitivism about normative judgment can be rendered consistent with the practicality of normative judgment and the Humean theory of motivation, then there can be no argument from the truth of the latter to the falsity of the former. Hence, the anti-realist must—if such a successful reconciliation can be had—critique the normative realist on other grounds. This ‘defanging’ of one of the anti-realist’s principal line of argument against realism therefore constitutes a significant virtue of a reconciliation of the propositions making up the practical problem.

In the rest of this paper, I develop and defend such a reconciliation. The normative realist can, I claim, reconcile the proposition that normative judgments are beliefs with the practicality of normative judgment and the Humean theory of motivation, thereby solving the practical problem. The (low) price to pay for this solution is the acceptance of a strong form of cognitivism about intention, and the making of a small—yet highly plausible—adjustment to our theory of what it is to be a motivational state.

But, as we saw above, the propositions (1) to (4) making up the practical problem are logically inconsistent. How then is a reconciliation possible? To mollify this worry, I should say in advance that my reconciliation will proceed through an adjustment to the definition of a motivational state in (MOTIVATIONAL). Once we accept that what it is to be a motivational state is not to be such that, taken together with some certain background belief, it can causally suffice for action, but rather to be such that it can causally suffice, (perhaps) in tandem with some certain background belief, for an intention to act, the inconsistency between (1) to (4) disappears. The way is then open for the advertised reconciliation.

3 Cognitivism about intention

Before I explain my solution, I should first say something about the details of the philosophical psychology that it presupposes. Cognitivism about intention is the view that your intentions to act are beliefs about what you are going to do. In its strongest formulation your intention to F is identified with, or wholly reduced to, the belief that you will do F. This is now a prominent view in the literature. Philosophers who have defended cognitivism about intention, in this strongest form, include Velleman (1985, 1989), Setiya (2007), Ross (2009), and, most recently, Marusic and Schwenkler (2018).

How do these cognitivists propose that we make sense of intentions as a species of belief? The standard view in philosophical psychology on the nature of intention is best represented by the work of Bratman (1987). For Bratman, intentions are commitments to action. In the same way as beliefs are theoretical commitments to the truth of a proposition, intentions are practical commitments to taking some course of action. Bratman and other proponents of the standard view further hold that intentions are sui generis mental states, on a par with belief and desire, and irreducible to them. Intentions, as Paul (2009)—another proponent of the standard view—puts it, are ‘distinctive practical attitudes’. Nevertheless, the standard view still has it that we can characterize intentions and their status as commitments to action. Most pertinently for us, intentions are held to be (1) conduct-controlling and (2) subject to certain norms of practical rationality. So, for cognitivism about intention to constitute an adequate theory of intention, its proponents must be able to accommodate these common ground desiderata.

How can cognitivists accommodate these elements of the common ground on the nature of intention? Consider, first, the fact that intentions are conduct-controlling—that is, the fact that your intention to F is disposed to prompt you, at the appropriate time by your lights, to actually do F (or, at least, to attempt to do F). How can mere beliefs about what you are going to do be conduct-controlling? How can they causally suffice for action? And, if the cognitivist simply stipulates that such beliefs are conduct-controlling, then isn’t she committed to denying the Humean theory of motivation? After all, if a belief about what you are going to do can causally suffice, by itself, for action, then it is manifestly not the case that action is only ever brought about by a belief and a desire working in tandem, contra the Humean theory.

To answer the last question first: no—the cognitivist about intention is not committed to denying the Humean theory of motivation, not even as an upshot of rendering her candidate beliefs for intentions conduct-controlling. On the contrary, not only is cognitivism consistent with the Humean theory of motivation, most cognitivists about intention actually endorse it as a part of their explanation of how their intentions are conduct-controlling. Cognitivists about intention normally seek to explain how the beliefs about your future actions that constitute, on their psychology, your intentions to act are conduct-controlling through reference to a background desire: for example, the desire to act as you expect yourself to act. Per the Humean theory of motivation, your belief that you are going to do F can combine with your desire to act as you expect yourself to act to cause you to actually do F. This, or some other, ‘background condition’ desire is standardly theorized by cognitivists to be present in all possible agents in virtue of their very nature as agents. In this way then, cognitivists about intention can, wholly in accord with the Humean theory of motivation, explain how the beliefs that constitute, on their view, intentions to act count as conduct-controlling (Velleman 1989; Setiya 2008; Ross 2009).

Let us call this strategy for explaining how beliefs about future actions can be conduct-controlling ‘Humean cognitivism about intention’. As I said above, Humean cognitivism about intention is not some fringe version of cognitivism, but actually the dominant variant. One concrete example of such an account is David Velleman’s (1985, 1989) cognitivist theory of intention. For Velleman, your intentions to act are certain beliefs about what you are going to do. Per Humean cognitivism about intention, Velleman holds that such mental states are only conduct-controlling in tandem with a certain ‘background condition’ desire, one that is constitutive of (rational) agency. For Velleman, this is your desire for self-knowledge—that is, your desire to know what you are doing when you are doing it. According to Velleman, you can satisfy your desire for self-knowledge by only doing what you predict that you’re going to do. After all, if you only perform actions that you expect yourself to perform then, not only will you never be surprised by your own actions, but you will be in a position to always know what you are doing while you are doing it. Consequently, your beliefs about what you are going to do will combine with this desire for self-knowledge to make you act as you predict you will act. Velleman further holds that this desire is constitutive of rational agency: all rational agents, by their very nature, desire to know what they are doing when they are doing it.

Other Humean cognitivists about intention include Setiya (2008) and Ross (2009). Given the plausibility of the Humean theory of motivation, these cognitivists also appeal to some ‘background condition’ desire—that is theorized to be present in all possible agents in virtue of their very nature—to explain how the beliefs that they claim constitute intentions are conduct-controlling. However, unlike Velleman, both Setiya and Ross are non-committal on which desire exactly would serve as this ‘background condition’ desire. But they are both convinced that such a desire is needed and that it must be present in all possible agents.

We have now seen how cognitivists about intention can explain our first desiderata on any adequate theory of intention, the fact that intentions are conduct-controlling. Let us now turn to the second desiderata: the fact that intentions are subject to certain practical norms. For example, in light of the fact that I have decisive reason to entertain my guests at my party, I am rationally required—or ought—to form the intention to do just that. This contrasts with the case of belief. Beliefs are standardly thought to be subject only to epistemic norms (Adler 2002; Shah 2006; Parfit 2011; Way 2016). The belief that p, according to the standard view, is governed only by evidence that speaks for or against the truth of the proposition that p—and not by any practical considerations concerning whether or not p is a good state of affairs, or whether there is practical reason to bring it about that p, or even whether you have good practical reason to make it the case that you believe that p. For example, in light of decisive evidence from hard experience that I will unintentionally offend my guests with one of my jokes, I am rationally required to believe that I will offend them, even though—let us suppose—I have plenty of practical reason to try and avoid doing so.

This doctrine—that beliefs are subject to epistemic norms alone—is known as ‘evidentialism about belief’. It poses an obstacle to cognitivism about intention. After all, if intentions are governed by practical norms, but beliefs—including beliefs about what you are going to do—are subject only to epistemic norms, then your intentions to act, contra cognitivism about intention, cannot be identified with, or reduced to, beliefs about what you are going to do. It should be noted—in contrast—that the fact that beliefs are subject to epistemic norms is no objection to cognitivism about intention. On the contrary, it is one of the principal explanatory virtues of cognitivism: by reducing intentions to certain beliefs the cognitivist can elegantly explain some of the norms of practical rationality governing intention—such as norms of consistency (with other intentions and background beliefs) and norms of means-end coherence, which both have an ‘epistemic character’—in terms of norms of epistemic rationality governing belief (Harman 1976; Setiya 2007; Ross 2009). But the fact, on the face of it, that intentions, but not any beliefs, are subject to practical norms does look to pose a potentially fatal objection to the cognitivist’s project.

How can cognitivists respond here? One popular move, originating with Harman (1976) and defended most recently by Marusic and Schwenkler (2018), is to simply stipulate that the beliefs about your future actions that constitute, for the cognitivist, your intentions to act are governed by the practical norms in question (in addition, of course, to the epistemic norms to which such beliefs are manifestly subject). After all, if intentions, by their very nature, are subject to certain practical norms, then, if intentions are really certain beliefs, then these beliefs must likewise be subject to those practical norms. These cognitivists insist that, in light of the fact that you have decisive reason to F, you are warranted in forming the belief that you will do F—so long as this belief is a self-fulfilling conduct-controlling mental state that will prompt you, at the right time by your lights, to actually do F. In this way then, cognitivists about intention deny evidentialism about belief, the doctrine that belief is governed by epistemic norms alone, in favor of a form of pragmatism, the view that (certain) beliefs are governed in addition by practical norms (Shah 2006; Parfit 2011; Way 2016).

A variation on this theme is articulated by Velleman. He agrees that the beliefs about what you are going to do that constitute your intentions to act are warranted in light of practical considerations. As he puts it ‘…they have been adopted by the agent out of a desire for their fulfillment…’ (Velleman 1989). What is distinctive about Velleman’s view, however, is how the formation of these beliefs is conceived by the agent as a means of making themselves act in the desired way. Again in his words, such an agent ‘…forms the conscious expectation of performing the action, in order to prompt himself to perform it….’ (Velleman 1989). For Velleman then, you form the belief that you will do F in light of your desire that you do F—or your belief that you ought to F—and your belief that you can bring it about that you do F by forming the belief that you will do F. This newly formed belief that you will do F then combines, on his picture, with your desire for self-knowledge to move you to actually do F. In this way too then, the cognitivist about intention can hold that the formation of certain beliefs about you are going to do—those self-fulfilling ones that constitute your intentions—can be warranted, even in the absence of sufficient evidence to form them, in light of practical considerations alone.

Cognitivists’ embrace of a limited form of pragmatism about belief also allows them to elegantly solve a looming problem for their view—namely, whether they can account for the commonsense distinction between intending to do something and merely foreseeing that you will do it. After all, for cognitivists, both your intention to F and your mere foresight that you will do F are instances of the belief that you will do F. But clearly there is a difference between intending to do something and merely foreseeing that you will do it. For example, I might foresee, in light of hard experience, that I am going to (unintentionally) offend my guests at my party. But I do not thereby count as intending to offend them. In fact, in light of my reasons to avoid upsetting anyone, I intend to try and avoid offending them. So if intentions are just beliefs about what you are going to do, as the cognitivist maintains, then the cognitivist owes us an explanation of the evident distinction between intending to do something and merely foreseeing that you will do it.

Our cognitivists, who have now embraced a species of pragmatism about belief, can now explain this distinction between intention and mere foresight through reference to a distinctively practical genealogy. In general, the belief that you will F counts as your intention to F just when, and because, you hold that belief in light of practical reasoning; otherwise, it counts as mere foresight. As Marusic and Schwenkler put it: ‘…intentions are beliefs—beliefs that are held in light of, and made rational by, practical reasoning….’ (2018; 1). So beliefs about what you are going to do count as foresight, on this account, when they are held purely in light of evidence; whereas such beliefs count instead as intentions just when they are held in light of, and made rational by, practical reasoning—that is, process of weighing the considerations for and against some course of action in light of your desires or your believed reasons for action etc. In this way then, through appeal to a practical genealogy, the cognitivist can accommodate the commonsense distinction between intention and foresight.

A seemingly problematic upshot of our cognitivists’ stipulation—that beliefs can be governed by the very same practical norms to which intention is agreed to be subject—is that her theory of the mind now licenses of a form of wishful thinking (Paul 2009). After all, if you are warranted, in light of your belief that you ought to do F, in forming the belief that you will do F, even in the absence of sufficient evidence to form this belief, then you are justified in believing that some state of affairs will obtain merely because you think that it ought to obtain. And that is just what it is to be engaged in wishful thinking. But wishful thinking, it is generally agreed, it a form of epistemic irrationality: it is irrational to believe that something will occur, simply because you think it would be better if it did or because you believe it should happen. Rather, you are only warranted in forming any arbitrary belief, it seems, if you have sufficient evidence—or epistemic reason—to warrant you in believing it. Belief, after all, aims at the truth. To act otherwise is to violate the tenets of epistemic rationality.

It is indisputable, I think, that this form of cognitivism about intention licenses wishful thinking. But what is contestable, however, is whether these instances of wishful thinking are really prohibited by epistemic rationality. Velleman has argued—convincingly, in my view—that the formation, in the absence of sufficient evidence, of beliefs about what you are going to do is not forbidden by epistemic rationality, on the cognitivist’s psychology, because these beliefs are self-fulfilling—that is, they typically bring about the very action that they represent—and are represented by the subject as being self-fulfilling. Consequently, on his account, your believing that you will do F gives you sufficient epistemic reason to believe that you will do F. Hence, such beliefs will be epistemically warranted and consistent with the requirements of epistemic rationality once they are in place. Wishful thinking, Velleman concludes, is therefore consistent with the norms of epistemic rationality—so long, that is, as the beliefs formed are self-fulfilling (Velleman 1985, 1989). In this way then, the cognitivist about intention can defuse the objection from wishful thinking.

I will now finish this overview of how cognitivism about intention can accommodate the basic desiderata on any adequate theory of intention with a brief review of its principal theoretical virtues. As I noted before, if your intentions are really beliefs about what you are going to do, then we have elegant explanations of some of the norms of practical rationality that govern intention—such as norms of consistency and of means-end coherence—in terms of the norms of epistemic rationality to which belief is subject (Setiya 2007; Ross 2009). We also get neat explanations of how—for example—you have non-observational (or ‘spontaneous’) knowledge of what you are doing while you are doing it (Velleman 1989; Setiya 2008), of why verbal expressions of intention—‘I am going to the park’—sound just like verbal expressions of belief (Marusic and Schwenkler 2018), and of why settling the practical matter of whether or not to do F normally settles (for you) the theoretical question of whether or not you are going to do F (Moran 2001; Marusic and Schwenkler 2018). Considerations such as these, and others, have been judged by many philosophers to give us compelling reasons to accept cognitivism about intention in one form or another.

4 A solution to the practical problem

The pieces are now all in place for me to outline the advertised solution to the practical problem. Recall that an adequate solution to the practical problem—one that reconciles the propositions making it up—will allow the normative realist to hold that normative judgments are both beliefs and motivational states, all consistent with the Humean theory of motivation. Here I propose that the normative realist can affirm all of these doctrines just by embracing cognitivism about intention and by making a small adjustment to our theory of what it is to be a motivational state.

How does my solution go? Let’s begin by assuming the truth of cognitivism about intention. As we saw above, cognitivists about intention hold, not only that your intentions to act are really beliefs about what you are going to do, but also that these beliefs are governed by both epistemic and practical norms—in particular, by the practical considerations that everyone agrees govern intention. Let’s further suppose that your normative judgment that you ought to F is just your belief that you ought to F. Now, since your self-fulfilling beliefs about your future actions are subject to those practical norms that govern intention, you are rationally required to form, in light of your belief that you ought to F, the belief that you will do F. Therefore, your belief that you ought to F will cause you, if rational, to form the belief that you will do F. (Perhaps, per Velleman’s psychology, in tandem with your belief that you can make yourself do F by forming the self-fulfilling belief that you will do F). Of course, the belief that you will do F, for the cognitivist, is your intention to F. It will cause you, in concert with your background desire to act as you expect yourself to act (or, for Velleman, your desire for self-knowledge), to actually do F. In this way then, your normative judgments can prompt you to act in the way that they indicate you ought to act.

So that’s my story. It allows the realist to solve the practical problem by allowing her to reconcile the propositions making it up. I now detail how it does just that. The first such proposition is cognitivism about normative judgment, the doctrine that normative judgments are beliefs—in particular, that your judgment that you ought to do F is just your belief that you ought to do F. As we saw at the outset, the normative realist is committed to this doctrine by her normative metaphysics. The proponent of my solution can affirm that your normative judgments are beliefs. For example, she can hold that, in light of decisive evidence that you ought not to F, that you would, if rational, revise your judgment that you ought to F.

The second such proposition is the practicality of normative judgment, the doctrine that normative judgments are motivational states. My solution allows the normative realist to hold that her normative judgments are motivational states. How so? Well, your belief that you ought to F can—perhaps in tandem with your belief that you can prompt yourself to F by forming the belief that you will F—cause you to form the belief that you will do F. And this belief, for the cognitivist, just is your intention to F. It will move you—in concert with some certain background condition desire—to do F. But how does this pertain to the matter of your normative judgment being a motivational state? Well, as we just saw, your belief that you ought to F can—perhaps in conjunction with some certain background belief—cause you, if rational, to form the intention to F. This is just like how desires combine with background means-end beliefs to produce intentions to act. Very plausibly, I want to say, a mental state possessing this property of being causally sufficient, in a rational agent, to produce—perhaps in tandem with a background belief—an intention to act is sufficient for that mental state being a motivational state. After all, that is plausibly why desires count as motivational states. Desires, on many views (e.g. Bratman 1987), do not directly cause overt behavior: their influence is always mediated by intervening intentions. Hence, it is plausible that what does make many desires count as motivational is the fact that they can combine with the right background beliefs to generate an intention, and not the fact that they can combine the right background beliefs to directly produce action.

(MOTIVATIONAL2) mental state M is a motivational state =df M can—perhaps in tandem with a belief—causally suffice, in a rational agent, for an intention to actFootnote 6

Granting this definition of a motivational state, your belief that you ought to F counts as motivational state. It causally suffices, in a rational agent, perhaps in conjunction with a certain background belief, for the formation of an intention to act. In this way then, the realist can hold that your normative judgment that you ought to F is both a belief and a motivational state, simply by assuming a strong form of cognitivism about intention and by making a small (yet highly plausible) adjustment to our theory of what it is to be a motivational state.

So far, so good. But what about the Humean theory of motivation? How can the proponent of my reconciliation affirm that doctrine as well? As I noted before, one might be initially inclined to think that the cognitivist about intention is committed to rejecting the Humean doctrine that action is only ever brought about by a belief and a desire working in tandem. But, as we just saw, this is not the case: cognitivists about intention typically seek to explain how your beliefs about what you are going to do can be conduct-controlling through appeal to a background condition desire—say, to act you expect yourself to act. This is wholly in accord with the Humean theory of motivation. Consequently, the normative realist proponent of my solution to the practical problem can affirm this element of Humean psychology: action is only ever brought about a belief and desire working together.

But, you might protest, my reconciliation still violates Humean psychology under another interpretation, according to which intentions to act are only ever brought about by a belief and desire working together (Arruda 2016). After all, on my theory, your normative judgment is a belief that suffices, perhaps in conjunction with a certain background belief, to cause you, if rational, to form an intention to act, regardless of whatever your desires may be. This directly contradicts the Humean theory of motivation, understood in this second way.

The best response I think I can make here is to simply concede the point: beliefs alone can causally suffice for intentions. Your belief that you ought to F—certainly in conjunction with your belief that you can prompt yourself to F by forming the belief that you will do F—can causally suffice for you, if rational, to form the intention to F. However, I think I can ameliorate this concession by observing that this intention is really just one of your beliefs about what you are going to do. And there is nothing mysterious—or ‘un-Humean’—about beliefs causally sufficing for the formation of further beliefs. For the cognitivist about intention, your belief that you ought to F can cause you, if rational, to form the belief that you will do F, in the same way as your belief that you ought to believe that p can cause you, if rational, to form the belief that p. There is nothing worryingly ‘un-Humean’ about any of this. The Humean theory of motivation issues no requirements or prohibitions on how beliefs interact with other beliefs—it is silent on the matter.

Of course, one might protest that there is something ‘un-Humean’ about this concession—namely, that it violates the Humean theory of motivation when understood in Arruda’s (2016) way. After all, for Arruda, there is at least one case in which beliefs alone causing sufficing for the formation of another belief is ‘un-Humean’: when that latter belief constitutes an intention to act. In such a case, beliefs alone will be causally sufficing for the formation of an intention to act, contra Arruda’s interpretation of the Humean theory of motivation.

However, I am quite sanguine about my solution to the practical problem violating Humean psychology, understood in this latter way, since my solution preserves it when interpreted in the standard way. The standard interpretation of the Humean theory of motivation, as we saw, commits it only to a prohibition on beliefs alone causally sufficing for action, and not to a prohibition on beliefs alone causally sufficing for the formation of an intention to act. Most pertinently, this standard formulation is the only one that the practical problem presupposes. Given all this, I am content with my solution to the practical problem only preserving the Humean theory of motivation under this standard interpretation.

The proponent of my solution can also affirm the other element of Humean psychology upon which the practical problem rests: the proposition that no belief is a desire. One might worry that my purported reconciliation is committed to denying this. After all, on my account, beliefs about what you ought to do occupy the functional role of a motivational state: they are apt to causally suffice, perhaps in concert with a certain background belief, for the formation, in a rational agent, of an intention to act. And that, I claimed, it just what it is to be a motivational state. Isn’t this sufficient for these beliefs being desires? Just like a desire, they can combine with a background belief to bring about an intention to act.

In brief: no. Being a motivational state, in the defined sense, is not sufficient for being a desire. There is more to being a desire, I think, than being a motivational state. For example, unlike beliefs, desires are not subject to the evidence (Hume 1739; Schroeder 2007).Footnote 7 In light of decisive evidence that p is not the case, you ought to revise your belief that p, but not your desire that p. But the normative realist’s candidate for a normative judgment, your belief that you ought to F, is subject to the evidence: in light of decisive evidence that you ought not to F, you are rationally required to revise your belief that you ought to F. Hence, for the proponent of my solution to the practical problem, your belief that you ought to F is a motivational state that is not a desire. In the absence of any other beliefs that threaten to also constitute desires, the advocate of my reconciliation can affirm the proposition that no belief is a desire, the second element of the Humean theory of motivation upon which the practical problems turns.

We are now in a position to understand how my solution defuses the practical problem for the normative realist. As we saw, the practical problem can be posed as a quartet of individually plausible but jointly inconsistent propositions: that normative judgments are beliefs; that normative judgments are motivational states; that action is only brought about by a desire and a belief working in tandem; and that no belief is a desire. The inconsistency here—granting that what it is for some mental state to be a motivational state is for it to be such that it can combine with a belief to cause action—goes like this: if a normative judgment is a belief, but not a desire, and such that it can combine with a belief, that is not itself a desire, to cause action then beliefs alone can cause action, contra the Humean claim that only a belief and a desire together can cause action.

My solution, I claimed, allows the realist to reconcile these seemingly inconsistent propositions. How so? Well, as we just saw, my machinery allows the realist to hold that her normative judgments are both beliefs and motivational states, all consistent with action only ever being brought about by a belief and a desire working in tandem and with no belief being a desire. The above inconsistency is removed through adjusting the definition of a motivational state. As we saw above, it is highly plausible that what it is for a mental state to be a motivational state is for it to be such that it can combine with a belief to cause an intention to act, not to directly cause action. Now that we are understanding motivation in this way, the inconsistency disappears.

5 Objections defused

So that’s how I propose that the normative realist can solve the practical problem. I’m now going to try to defuse a few of the objections that I anticipate to this account.

The first objection goes like this: how does this theory of normative judgment differ from externalism? (Externalism, recall, is the view that normative judgments are not motivational states). After all, on both stories, normative judgments can bring about action only if some ‘background condition’ desire is present. On the account at hand, your belief that you ought to F only prompts you to actually do F through first causing you, perhaps in conjunction with your belief that you can prompt yourself to do F by forming the belief that you will do F, to form the belief that you will do F, a belief that only then only causes you to do F through combining with a background desire—for Velleman, for example, your desire for self-knowledge. And, likewise for the externalist, your belief that you ought to F can only move you to actually do F through combining with some certain background desire—for example, your desire to do whatever you ought to do or your desire to be practically rational etc. Given this, why shouldn’t we think that the story being told here is not simply some complicated version of externalism?

Well, my account has it that normative judgments are motivational states. Externalists, in contrast, hold that normative judgments are not motivational states. After all, on my account, your belief that you ought to F—that constitutes your normative judgment—counts as being a motivational state because it can causally suffice, perhaps in combination with certain background beliefs, for you to form an intention to act. But, for the externalist, your belief that you ought to F does not count as a motivational state. Rather, it causes action, in accord with the Humean theory of motivation, only through combining with a certain background desire—your desire to do whatever you ought to do, say. So, whereas my account seeks to accommodate and explain the practicality of normative judgment, externalists simply deny it. This is why my account is not just another variation on an externalist theory of normative judgment.

This response gives rise to a second related objection: since both accounts are committed to the existence of some background desire to explain how normative judgments can cause action, what dialectical advantage does my story hold over externalism? After all, as we just saw, externalists are free as well to posit a background desire to explain how normative judgments can motivate action. Why prefer my machinery?

I have two observations to make in response here: first, unlike any version of externalism, my machinery preserves, and explains, the practicality of normative judgment. As we saw before, in the vignette with which I began this paper, there is good evidence to believe that normative judgments are motivational states: they play the role of a motivational state in our commonsense psychological explanations of behavior (Shafer-Landau 2003). Our evidence suggests, not merely that normative judgments play some role—such as the role any belief might play—in the motivation of action, but rather that they occupy the role of a motivational state (such as a desire). Externalists have normative judgments playing the ‘belief-role’ in the explanation of action, whereas my account has them playing (something similar to) the ‘desire-role’. My machinery is therefore to be preferred since, unlike externalism, it can accommodate the practicality of normative judgment, whilst also accounting for all the other relevant desiderata.

Second, my theory also enjoys a dialectical advantage when it comes to accounting for the fact that normative judgments seem to be capable of motivating any possible agent who cognizes them to action. For the externalist, your normative judgments will only be capable of motivating action in conjunction with your desire to do whatever you ought to do (or whatever). Since it appears to be perfectly possible for an agent to lack this desire, the externalist consequently cannot account for the fact that normative judgments are capable of motivating any possible agent who cognizes them to action. The externalist must either insist (against the appearances) that such a desire is present in all possible agents capable of normative judgment or, alternately, hold that there are possible agents whose normative judgments are incapable of moving them to act since they lack the requisite background desire.Footnote 8

But surely, the reader might protest, my machinery is no better off in this respect? After all, it seems possible that there could be an agent who lacked the desire to act as they expect to act or the desire for self-knowledge (or whatever the cognitivist about intention’s preferred ‘background condition’ desire might be). The proponent of my theory must therefore either insist (against the appearances) that such a desire is present in all possible agents capable of normative judgment or, alternately, hold that there are possible agents whose normative judgments are incapable of moving them to act since they lack the requisite background desire. (Indeed, such agents would be incapable of action at all since intentions, for the cognitivist, only produce action at all in tandem with such ‘background condition’ desires).

However, all is not equal, dialectically speaking, between the externalist and the proponent of my machinery on this point. Cognitivists about intention have already invested much energy in justifying their view that their preferred ‘background condition’ desire is constitutive of (rational) agency and consequently present in all possible agents capable of normative judgment (See, for example, Velleman (1985, 1989) for an extended defense of his view that the desire for self-knowledge is part of what it is to be a rational agent). The proponent of my theory of normative judgment can simply help herself to this story since she is helping herself to cognitivism about intention. No externalist—to my knowledge—has ever defended the view that the desire to do whatever you ought to do, or the desire to be practically rational, is present in every possible agent capable of normative judgment.Footnote 9 In advance of that project being convincingly pulled off, the dialectical high ground on this point is occupied by the proponent of my machinery.

I want to consider one last objection here that concerns the motivational power of normative judgments: it is natural to think that your normative judgment that you ought to F provides you with some degree of motivation to F, even in those circumstances in which you are akratic and end up forming the intention to perform some incompatible action. But it is not obvious that my machinery can account for this. After all, my account has focused on explaining how normative judgments can be conduct-controlling in a practically rational agent.

However, I think that I can quite naturally accommodate the phenomenon of a normative judgment providing you with some degree of motivation to act even when it is not conduct-controlling. How so? Well, simply in virtue of desiring to F you count as being motivated to some degree to do F—in proportion to the degree to which you desire to F. More generally, if you have some motivational state to do F, then you count as being motivated to some degree to do F in virtue of being in that motivational state. Since, on my account, your belief that you ought to F counts as a motivational state to F you therefore count, on my view, as being motivated to some degree to do F simply in virtue of being in that state, even if this state is not conduct-controlling. Of course, you would also count as being akratic, and by your own lights, in such a circumstance, since you would believe that you ought to F yet would have failed to form the intention to F. But this is wholly consistent with you counting as being somewhat motivated to F.

Lastly, there are normative judgments beyond your judgment that you ought to F. For example, your judgment that you have some reason—though insufficient—to F. Can the proponent of my machinery accommodate the fact that such judgments would provide a rational agent with some motivation—though, of course, non-overriding—to F? I think so. It is plausible that the degree of motivation to F that some mental state M provides you with, if rational, is determined by the degree of rational pressure M exerts on you to form the intention to F. Since, on my account, your belief that you have some reason—though insufficient—to F exerts some (non-decisive) rational pressure on you to form the intention to F—that is, for the cognitivist, a belief that you will F—it therefore counts as providing you with some (non-overriding) motivation to F. In this way then, the proponent of my solution to the practical problem can accommodate the commonsense thought that normative judgments, even when not conduct-controlling, provide you with some degree of motivation to act as they indicate that you should, or as you have reason to, act.

6 Conclusion

This completes my development and defense of the advertised solution to the practical problem. Normative realism faces a problem concerning the practicality of normative judgment, the presumptive view that normative judgments are motivational states. But normative judgments, for the realist, must be beliefs. This seemed problematic because it is difficult to see how any belief could have the essential connection to motivation necessary to account for the practicality of normative judgment. Here I showed how the realist can hold that normative judgments are both beliefs and motivational states, all consistent with the Humean theory of motivation. The (low) price to pay for this reconciliation is the acceptance of a strong form of cognitivism about intention and a small—yet highly plausible—adjustment to our theory of what it is to be a motivational state.