Reason in Hume's Passions

Abstract

Hume is famous for the view that "reason is, and ought only to be, the slave of the passions." His claim that "we are no sooner acquainted with the impossibility of satisfying any desire, than the desire itself vanishes" is less well known. Each seems, in opposite ways, shocking to common sense. This paper explores the latter claim, looking for its source in Hume's account of the passions and exploring its compatibility with his associationist psychology. We are led to the conclusion that this view—that desires vanish when fulfilment is deemed impossible—endows reason with a power over the passions that is at odds with its role as slave, and ultimately incompatible with a proper understanding of emotions such as grief. Such emotions involve continuing to want what one believes to be impossible. The human (and Humean) imagination can sustain desires without the belief that fulfilment is possible.

[N]othing is more certain, than that despair has almost the same effect upon us with enjoyment, and that we are no sooner acquainted with the impossibility of satisfying any desire, than the desire itself vanishes.

T Intro. 9; SBN xviii

Introduction

Why is Hume certain about this claim?

Hume famously says that "reason is, and ought only to be, the slave of the passions and can never pretend to any other office than to serve and obey them" (T 2.3.3.4; SBN 414).² He proceeds to argue that reason is inert, that the truth or falsity of beliefs about the world cannot, by themselves, move us to act without the contributions of the passions. Kant's practical reason is not inert in the same way. The Categorical Imperative is capable of constraining us in the pursuit of ends that the inclinations (passions) present as possibilities. This represents a familiar contrast between the two philosophers. But recent work on Hume has revealed that reason's "enslavement" to the passions is far from complete.³ There are many ways in which Humean reason exerts its influence on human conduct.⁴

In this essay we consider a passage in which Hume appears to go too far in the direction of rational control, attributing to reason a mastery of the passions that it does not seem to have. Interpreting Hume's famous discussion, "Of the influencing motives of the will," (where the famous reason/slave passage occurs) confronts one with the task of integrating the following (not so famous) line, which we denominate PYR (passions yield to reason):

PYR: The moment we perceive the falsehood of any supposition [that a passion is founded upon], or the insufficiency of any means, our passions yield to reason without any opposition.

(T 2.3.3.7; SBN 416–17)

On its face, PYR is as troubling as the claim that reason is the slave of the passions (and directly in conflict with it). Who has not continued to want things after discovering that they are impossible to obtain? Our paper explores Hume's theory, interpreting and assessing the plausibility of this view of reason's ability to control desire (in this way). We find several contexts in which this claim appears to be both Humean and plausible—though even these are in some ways problematic. We also find cases that do not yield to Hume's account. In this type of case a person continues to want what she knows she cannot have. This, in our view, is a central aspect of the passion we call grief.

1. The Problem

There are two points that Hume attempts to prove in defending the claim that reason is the slave of the passions: "That reason alone can never be the motive to any action of the will, and secondly, that it can never oppose passion in the direction of the will" (T 2.3.3.1; SBN 413). We will not question the first of these claims. The second, on the other hand (that reason "can never oppose passion in the direction of the will") appears to be at odds with the view that passions disappear when reason discovers that they are based on a false supposition, or discovers insufficient means to obtain that to which passion directs us. In fact, PYR seems to make us out to be extraordinarily rational in relation to our passions. If only we did stop wanting things as soon as we discovered "insufficient means," life would roll easy. What would the world be like if people never continued to want things that do not or cannot exist? Realizing I do not have enough money to buy something I really want does not cause my desire for it to disappear. Is not the traditional idea of a battle between passions and reason based on the very observation that such desires often do not always yield, even when our reason tells us that we cannot get what we want? If Hume is claiming that passion will always yield to reason in this kind of case his claim seems at odds with common sense.

2. The Power of Reason

But this is too quick and uncharitable. In order to understand Hume's point it is necessary to clarify the object of the passion he is referring to, for example, when he speaks of "the insufficiency of any means." What Hume means in saying "our passions yield to reason" is simply that we cease wanting to do something when we realize it is not a means to an end we want. If I desire to go to the café only to see a friend, then when I discover that my friend will not be there I stop wanting to go. Since my desire to go to the café was based on the mistaken belief that my friend was there, when reason tells me that I will not be able to achieve my end by going to the café, I no longer want to go. I might wish I could go see my friend at the café. But to actually go there (or even to want to) I would have to be confused (or have some other end in mind, such as having a drink). Purely instrumental desires do seem to yield to reason. When I discover I cannot raise the money for the car I want, I need not stop wanting the car, since this is not an instrumental desire. But, when I discover that my car-lottery ticket does not have the winning number, I drop it in the wastebasket, though a moment before I was treating it like a prized possession. Reason has shown me that I cannot win with this number, thereby breaking the link between the ticket and getting what I want. Of course, I can go on wanting the car itself. The end that I seek, it seems, is untouched by the discovery that what I had hoped was a means will not, in fact, work.

Perhaps this seems right. But let us consider it in relation to Hume's account of desire and aversion: "'Tis obvious, that when we have a prospect of pain or pleasure from any object, we feel a consequent emotion of aversion or propensity" (T 2.3.3.3; SBN 414). I see a rose and want to smell it. I smell a steak and want to eat. Afraid of the growling dog in my path, I feel like running. The prospect of pleasure from the rose or the food and the anticipation of pain from the dog's teeth are aspects of the emotive states that will move me to act—unless opposed by other affective states. Reason can, of course, be a factor in such motivation. When we want something, this very desire will also make us "cast our view on every side" (ibid.), in order to comprehend what is connected with the end desired by the relation of cause and effect. Fearing the running will incite the growling dog, I may constrain myself to walk slowly past. So, reason, as the source of causal information, plays a clear role in re-directing our passions and actions. (In fact, it is implicated in the very concept of desire, given that this involves the prospect of pleasure or pain; "prospect" is a causal term). In the last example, my desire to run does not move me; it is opposed by a desire to appear calm, which involves instrumental reasoning, but ultimately has its sources in the same aversion to being bitten.

What we want to emphasize about this case, however, is that losing the desire to run is by no means the only (psychologically) possible response. Most of us, in fact, will both think that it will make our situation worse (lessening the chance of escaping safely) and still feel a strong desire to run. Some of us will give in to that desire. This is a fairly clear instance of the problem of akrasia (prudential weakness). What is believed to be a clearly "insufficient means" to the end of escaping unharmed does not "yield to reason" as the PYR dictum requires.

Another potential problem with PYR might be predicated upon the above account of desire. Suppose I discover not just the insufficiency of a particular means of obtaining something I want, but the impossibility of obtaining it: there are no means sufficient to the end that I have. Thus, there is no "prospect of pleasure" from this object. If the lottery ticket were my last hope for the car, when reason extinguishes that hope, not just the ticket, but the car itself should cease be an object of desire, since a "prospect of pleasure" is part of what it is to desire something. Now, of course, this reversal might happen. We are familiar with the human tendency toward sour grapes, which reverses the attitude toward what is discovered to be unobtainable. But the "sour grapes" response is far from a universal outcome of finding that one cannot get what one wants. People often seem to go right on wanting what they believe they cannot have.

One might complain that this argument works by implausibly removing all hope of getting what is desired. Given the difficulty of eliminating every possible means of getting what one wants, we can go right on wanting things because some probability of getting them remains. But this is wrong; one often comes to believe that it is not possible to get what one continues to desire. When the "sold" sign goes up on the house that I want, when the taxi is miles from the airport at flight time, I realize that I cannot get what I continue to want. Human life is full of such frustrations—situations involving both the desire for something and the realization that getting it is impossible. PYR does not directly rule this out, as we saw at the outset of this discussion. But the causal anticipation that Hume builds into the concepts of desire and aversion does seem to rule out wanting what is believed to be impossible. As Hume says in the quote we have taken as our headnote, "we are no sooner acquainted with the impossibility of satisfying any desire, than the desire itself vanishes."

Again, one might object that the "prospect of pleasure" from an object is just a causal generalization of the form ‘X is pleasurable,' not the specific claim that one can (now or soon) get pleasure from X. So, believing that it would be fun to have a new car, I go right on wanting it, despite knowing I will not get it. This may seem right; but it does not fit Hume's account of reason as inert to suppose that this general claim could be the basis of a continuing desire. Why would the general knowledge that it is pleasant to have a new car sustain my desire in the face of a belief that it will not be mine? The Humean account does not seem to accommodate this relation between general beliefs and desires. Moreover, it is likely we all do in fact have beliefs about certain objects having the "prospect" of pleasure in this sense (X is pleasurable), without having a corresponding desire for the object. For instance, a belief that eating Christmas dinner with my family is pleasurable is not (at this moment at least) accompanied by the desire to eat Christmas dinner with them.

Paradoxically, Hume makes this alignment of (instrumental) reason and passion important to his argument that there can be no "opposition" between reason and the passions. He first allows that there are two ways in which passions can be said to be unreasonable (when we speak without precision). There are causal (means/end) judgments of the sort that we have just been considering, and there are also the cases in which "a passion such as hope or fear, grief or joy, despair or security, is founded on the supposition of the existence of objects that really don't exist" (T 2.3.3.6; SBN 415–16). Now, according to PYR, when the individual's reason reveals the falsehood of the supposition in question, the passion disappears. Strictly speaking, reason does not act in opposition to this passion, because the belief and the passion never co-exist. As soon as one ceases to believe that the object exists or comes to believe that her desire was predicated on a falsehood, the passion itself disappears. There is never any struggle ("opposition") between passion and reason because the unraveling of the relevant causal belief is at the same time the demise of the passion.

We can immediately point to an assumption in this argument that seems both non-Humean and dubious. This is the assumption that passions cannot be maintained by sources that lie outside of such founding beliefs, for example, by other passions. Off hand, it would seem that this is possible. But to determine whether it is, we need to reveal what Hume means when he speaks of beliefs that "accompany" passions and passions that are "founded on" belief. We will see that in some cases it is a necessary truth that a passion cannot continue, absent the associated belief. Losing one of the constitutive beliefs is necessarily ceasing to be in the relevant emotional state: that is, this is true by definition.

This, of course, is not Hume's own view of his account of the passions. But many commentators would agree with us that he gets his own theory wrong in this respect. Where he thinks he is giving the causal conditions of passions, he is in fact providing an account of their cognitive content. Annette Baier's account in A Progress of Sentiments provides an illuminating example.⁵ Somehow we must accommodate Hume's view that the passions are "secondary impressions" or "impressions of reflexion." A passion "would not be an impression ‘of reflexion' except for the fact that it has an idea to introduce it"; and of course the underlying idea does represent.⁶

3. The Logic of Passions

In fact, human (and Humean) passions are complex in their cognitive structure. An example that occupies a good deal of Hume's attention in Book 2 is pride. What is it to be proud of something? In part, it is to have certain positive feelings; but it could not be just this. There is, for one thing, no pride without beliefs about one's self and about some properties, possessions, or dispositions, that connect with oneself. A person can be proud of her home, her job, her ability to write. A smoker could be proud to have succeeded in quitting. Indirectly, one can also take pride in someone else's accomplishment. For example, I could be proud that you quit smoking; but only under special circumstances that connect you with me, for instance, I have been coaching you or you are my daughter. But, now let's consider what happens with the discovery that one of these "accompanying" beliefs proves to be false. I thought that it was my daughter who quit smoking, but my telephone evidence is faulty: it was her friend that she was talking about. In such cases, Hume's phrase seems to fit. My pride "immediately yields to reason." I cannot sustain my pride, when the belief it is "founded on" in this sense has gone. But here the explanation is not some revelation about what this does causally to my feelings. Pride itself is a mixture of beliefs (which can properly be governed by reason and evidence) and feelings (which, as Hume insists, cannot because they are not representational). Since the beliefs are partially constitutive of pride, it cannot be sustained when the beliefs are gone.⁷ In relation to some passions, then, it seems plausible to concede Hume's claim that our passions must yield to reason. Change of a material belief is not contingently connected with change of passion. It is impossible to imagine (to use Hume's test of relations of ideas) being proud of someone else's possessions without imagining as well that they connect with oneself, or imagining that one believes there is such a connection.

4. Frustration and Grief

We want to argue that in some cases our emotions depend on not being able to get what one continues to want. Consider grief over the loss imposed by a loved one's death. A belief in non-existence is itself partly constitutive of this passion. I might be saddened on reading an obituary for Bertrand Russell and this could properly be called grief. But upon reading Russell's own protest, the next day, that the "claims about my death are greatly exaggerated," I lose a belief that is a necessary condition of this passion. I could not possibly retain grief in the face of this belief change. Again we have an analytic truth. Notice that I might continue to feel sad or depressed, just as I might have trouble shaking off the depressive effects of a bad dream, even after I have realized that it was just a dream. What is necessarily true is that such residual sadness is not grief (in the relevant sense) once I have accepted that the person continues to exist.

In cases where grief is most manifest and painful, as in the loss of one's spouse or child, what is most painful is the realization that one can never again see or touch or talk (or interact in any of a huge variety of ways) with a person one has loved. These forms of interaction were sources of pleasure and security, and perhaps even sources of one's continuing identity—in a sense of "identity" that can be explained in Humean terms. Now, (it seems to us) if the desires to continue interacting in these ways did not persist, there would be no such thing as grief. It is partly because these passions do continue, and generally become greatly amplified, that grief exists. But, in this sort of case one also believes that there are no means that are sufficient for the satisfaction of the desires that form the affective background of grief. So, this appears to be a case that defies Hume's claim about the way in which our passions yield to reason (PYR). A person could not even experience the loss and frustration of grief if her desires conformed to the pattern that Hume describes in the Introduction to the Treatise where coming to accept "the impossibility of satisfying any desire," yields the result that "the desire itself vanishes" (T Intro. 9; SBN xviii). In this respect, Hume's account makes us out to be rational in the way exemplified by Star Trek's Mr. Spock. But this seems to miss an essential part of what makes us human.

Hume's own treatment of grief⁸ is instructive here. He classifies grief as a direct passion, that is, as one which does not (as pride and humility, love and hatred do) connect essentially with the idea of self or other. It is situated in another pair of pairs, viz., hope and fear, joy and grief. The first items in each of these pairs (hope and joy) are distinguished from the second (fear and grief) in terms of their hedonic tone: the objects of the former are perceived as good (pleasurable) and the later evil (painful) (T 2.3.9.8; SBN 439). What distinguishes hope from joy, and fear from grief is the certainty of the good or evil (T 2.3.9.6; SBN 439). Where we view something good as uncertain we experience hope; where we believe it to be certain or nearly so we experience joy. Likewise an uncertain evil is perceived as fear; where the evil is viewed as certain one experiences grief.

In this analysis Hume is using "grief" in a generic sense; not in the specific sense that we have been considering, the sense that connects with the death of a person with whom one has been closely associated. Thus, Hume speaks of anything both certain and bad as a source of "grief," only referring once (in the Treatise) to grief in relation to death.⁹ But, though he is interested in a wider array of phenomena, his own account of grief has elements in common with our own. The finality of death is one aspect of its certainty; there is no way back to the way things were. And, where there is uncertainty as to whether someone is dead (or will die soon) one experiences fear and anxiety rather than grief. It is also true that some of the other cases that we have discussed—missing one's flight, finding that one's dream house has already sold—could properly be occasions of "grief" in this wide sense. Where, then, is the dispute between Hume's account and our own?

It seems to be this. We have assumed that grief (whether it is for a person or a missed opportunity) always involves the consciousness of a loss. What one experiences as a loss is necessarily connected with what we have called the "affective background" of grief; that is, one continues to want what one has lost. From this we conclude that anyone who experiences grief must continue to want what she also knows she cannot have. (If there were doubt about this, one would experience fear—"I'm afraid I will miss the plane.") On the other hand, Hume identifies as grief the certainty of anything painful. This, because it is pain, will give rise to desires of various sorts. Being in pain, we will "cast our view on every side," to find an escape. An instrumentally rational mind would stop wanting what is impossible, stop dwelling on the past, as if it could be revised. But it is not an adequate account of grief to characterize grief as pain in something certain. Grief is the pain that involves a loss and it is experienced as longing for what one cannot have.

5. Reason's Influence on the Passions

Above we considered some of the ways in which, according to Hume, reason will influence what we want and what we end up doing. It will be worthwhile to summarize our account of this aspect of Hume's theory before going on to consider objections to our unorthodox account of the power over the passions that Hume attributes to reason. We have seen at least the following three sorts of powers of reason in Hume. Reason can

1. 1. Remove desires for a means by undermining the judgment that this is a means of satisfying the end desired. (This plane doesn't go to my destination. I don't want to get on.)

2. 2. Remove complex passions by undermining the judgments on which they are "founded" in a logical sense. (Not my house; it is impossible for me to be proud of it.)

3. 3. Remove desires for an end by undermining the judgment that there is any means of satisfying the desire. (The troubling cases of frustration: I want to be on that flight, but I can't make it; I want to see and talk with a person, but I know I can't - she's dead.)

These ways in which reason can limit what we desire grant reason some fairly hefty powers of directing and disabling our passions.

(1) The first claim seems right. Of course one does not want to continue on a path that one discovers is not headed toward the destination one seeks. Still, this view does not accommodate certain forms of "irrational" desire (e.g., to run from the dog, knowing this will make matters worse) or aversion such as phobias.⁹ We return to this problem below, in considering objections to our interpretation of the key passage (where the PYR dictum occurs).

(2) Hume can hardly be faulted for endorsing the second of the above ways in which reason limits what passion we have. At issue here are the complex passions that have implicit propositional (and hence representational) content. If one ceases to believe any of these presupposed judgments, then the emotion in question cannot continue to exist. As we have seen, one cannot take pride in something that does not connect with oneself; and one cannot—except in a nonliteral sense—"grieve" for a friend that one discovers to be living. Although Hume is not wrong about this, the full recognition of its implications would dramatically alter his theory of the passions. He would have to recognize, for example, that the neat division between reason and passion is itself problematic. It is problematic because reason will have the power to make discoveries that change beliefs in ways that belong to "the logic" of particular passions. It also makes problematic the claim (noted above) to the effect that reason cannot oppose passion because a passion "contains not any representative quality, which renders it a copy of any other existence" (T 2.3.3.5; SBN 415). Of course, a passion as a whole is not a representation of something else. But it is false that a passion "contains not any representative quality."

(3) It is the third implication of Hume's theory that seems to us to grant reason more control over the passions than is realistic. Here Hume seems, in fact, to identify human nature generally with a Stoic attitude toward what is unobtainable. Were we like this, we would never experience the sort of deep frustration that goes with the realization that one cannot get what one continues to want. Our reflections on grief have led us to the conclusion that in this respect Hume's instrumental rationalism grants too much power to reason. It seems clear that a grieving person cannot have lost the passions involved in caring for the person that has died.

We do not accept, as a plausible way out of this objection that the griever has not really adopted the belief in his loved one's death. Not only is this contrary to obvious interpretations of ways that such persons act—in preparing for a funeral, writing an obituary, and so on, it also conflicts with the very idea of grief, which presupposes the belief that one has lost the person for whom one grieves. Moreover, it is contrary to the certainty that on Hume's own view is a necessary condition of grief.

Another possible way out of the conclusion that people can want what they believe they cannot get is to employ the distinction between wishing and wanting: the grieving person merely wishes for what she believes to be impossible. Though this clearly deserves further consideration, our tentative response is to point to its likely emptiness. One might decide to call all situations in which people appear to want what they cannot get, cases of wishing rather than wanting. But, it seems unlikely that a person's emotional states (impressions of reflection)—and her pro-attitude toward doing various things such as talking with the person in question—have instantly shifted to some alternative attitude with the news that the person in question is dead. "Wishing" (in this context) just means wanting something that one also believes one cannot obtain. So, the new terminology just builds in the problem that we are faced with here. To us it seems better to bite this bullet directly. Understanding grief actually demands recognition that we sometimes face situations in which we continue to want what we know we cannot have.

6. Wanting the Impossible: Objections

Two central claims are being advanced in this paper. The first is that in the key texts concerning the way in which our passions "yield to reason," Hume characterizes reason as having a kind of dominance over the passions that is generally ignored. The second is that the control he asserts presents an implausibly rationalistic psychology that, in the end, will not accommodate widely held intuitions about grief and other forms of deep frustration that combine a desire for something with the realization that it is unobtainable. Obviously, doubts can be raised about each of these claims. It may be that we are not giving a plausible account of the key passages from Hume. Alternatively, or in addition, it may be that our assumptions about human psychology, and in particular of grief, are unsatisfactory. In this section we will consider each of these sources of doubt, concentrating primarily on the first.

(1) Interpretation: We need to return Hume's statement of PYR to its context. Hume has just raised (T 2.3.3.6; SBN 415–16) the question whether there is a sense in which passions can be said to be contrary to reason. The answer is that, in a loose sense, they can be contrary where the passion is founded on ("accompanied by") a belief that is false (an "object" that does not exist) or a faulty judgment of means to ends. The next passage is meant to illustrate the point:

I may desire any fruit as of an excellent relish; but whenever you convince me of my mistake, my longing ceases. I may will the performance of certain actions as means of obtaining any desir'd good; but as my willing of these actions is only secondary, and founded on the supposition, that they are causes of the propos'd effect; as soon as I discover the falsehood of that supposition, they must become indifferent to me.

(T 2.3.3.7; SBN 416–17)

One objection to our treatment of the key passage, then, is that Hume is making a more modest claim than we have supposed.¹⁰ If I only want this fruit because I think it is sweet—if I want it under precisely this description ("that sweet peach")—then it would seem that when reason removes support for the truth of this description, it must at the same time remove the desire for this peach. Moreover, Hume goes on to say, where something is wanted merely as a means to an end, change of belief that it is a means to that end must make it a matter of indifference. You can say that my desire to get on a plane is unreasonable, when you know that it does not go where I want to go. But when I myself discover that the accompanying belief is false, I must stop wanting to get on the plane. The claim is that we can desire an object only under a particular description; absent the description and the desire must be gone. Hume does not rule out our continuing to want a sweet peach after discovering that this one is not sweet, or even when I come to believe that it is too early in the season and there are not any sweet peaches. He rules out wanting this peach for its sweetness while at the same time believing it is not sweet. Nothing here (the objection continues) rules out our yearning for all sorts of things we know we cannot have, including the return of a person whom one knows to be dead.

Before responding to this important objection, we should make it clear that we are not deriving these problems about unsatisfiable desires from the key claim (PYR), taken by itself. We have assumed that desire and aversion are in some way connected with every passion that can move us to act (as all can—this is what makes them passions). We have based our case partly on the account of desire and aversion that Hume has offered (in this very section)—in particular, on the discovery that there is a causal judgment (prospect of pleasure or pain) built into every desire or aversion. If this is the case, then the realization that there are no means of satisfying a desire will be, as Hume says here, the end of that desire.¹¹

With this in the background, let us return to the key passage. It may seem obvious that one cannot go on wanting this peach (which one wanted only for its "relish") after one is convinced that it is not sweet. And, indeed, it is obvious that one cannot reasonably do so. But, of course, we are not concerned here with the mere tautology that it is unreasonable to want what reason excludes. The question is whether it is possible. We believe that this is possible, even in a case as simple as this. I may be convinced that the peach is not sweet. But, though only this "relish" is the causal foundation of my desire, I can continue to want it because it has properties—colour and texture for example—associated with sweetness that continue to motivate my desire. Here we are not saying that these properties maintain my belief in the sweetness of the fruit, since that would not give us the case that Hume offers. We are saying that the desire for something can be maintained by sources that do not themselves justify the belief about the object that was the original source of the desire. We will return to this point.

Our view may still be thought problematic in relation to Hume's theory of passions as "secondary impressions." Clearly there must be a primary impression or idea in order that any passion exist. In the case above, it is the peach, perceived as sweet, toward which one feels attracted; but when "you convince me of my mistake," my idea of the peach has changed. Thus, I have lost the object of my appetite.

Though this seems correct, it is really too simple to deal with the complexity of the human passions that Hume's theory must (and does) address. In fact, the imagination plays a huge role in the construction of the objects of our passions. Consider some different cases. A person sees a woman enter the café where he sits listening to music. Her dark hair, cut as it is, reminds him of a music teacher who was particularly kind and helpful to him. He responds positively to the woman who has just arrived, as if she too were kind. But he is not naive enough to actually adopt this belief about her. Another person reads about a young boy suffering from cancer of a type that is not currently treatable. Immediately she begins to long for a cure, though she believes that none exists. She is moved by this desire to make a contribution to the Sick Children's Fund. The image of the boy's face is still with her when she later decides to go into the study of medicine. In these cases, as elsewhere, passions have a life of their own, governed by processes of imagination not necessarily constrained by processes of reason. We want all sorts of things that we believe not to exist. In some respects whatever we can want, must be believed not to exist, given the direction of fit between desires and the world. Hume's theory of the imagination gives an account of the construction of the ideas that become the objects of our desires. Some of these processes of construction are built around associations that form our causal beliefs about the world (which also involves projection from what has been observed to what has not). But not all of them are constructed or maintained by such rational processes of causal association. It is to be expected that the Humean mind will often contain both ideas and emotions that are not hostage to our considered (rational) beliefs about the world.

It might be thought that when I want the fruit because of its sweetness, it is really just this property that is the object of my desire, not the peach taken as whole. Our argument (here) turns in part on the possibility that the desire for the peach is sustained by properties associated with sweetness. It might also be thought that these are separate desires (e.g., for a fruit of a particular colour) and hence irrelevant to the case in question, where only the "relish" of the fruit is in question. But, this too oversimplifies the idea of the object of an affective state. When one wants the peach for its flavour, one does not just want its particular sweetness. One wants this together with the other qualities which, taken together, constitute it as an object. Similarly, in the above example it is the woman (not simply her hair) that is the object of positive feelings, though the cut of her hair that is their source. The attraction may even survive the loss of that property when one has a more accurate view, because the attraction comes to be carried by other associated properties.¹²

To understand this we must see that the imagination plays a key role in relation to the passions. In the passage quoted above, Hume's argument assumes that belief is centrally involved in the maintenance of the passions, including this simple desire for the object of a particular appetite. Now, as we have seen, there is a way in which causal associations are implicated in desires, since every desire embeds a prospect of pleasure. But are the objects of desire always constrained by our beliefs about the way the world now is? Can processes of imagination that are in conflict with a person's beliefs sustain desires? There is nothing in the Humean account of desire that rules this out. It is true that desires, like all passions, are impressions of reflection. This means that they are always supervenient upon some idea.¹³ But, imaginative projection is a normal source of the ideas that are the basis of passions. In fact, associative projections of the imagination always carry us beyond what is merely perceived or remembered, when desires are formed. This car is not mine; but I can imagine it as mine and even imaginatively feel the pride that I would take in it. I imagine the red apple I see as tasting juicy and sweet.

In the normal case, reason-based beliefs will constrain the processes of imagination that engender and sustain beliefs. If you convince me that the peach is not sweet, I will no longer want it. But this is a contingent matter. The imagination, which in relation to desire, always leads us beyond what we believe to be the case in some respects, can maintain desire in the face of belief. I can (in Davidson's example) yearn to drink a can of paint, despite my conflicting belief that it would be most unpleasant! It is true as well, of course, that desire is always relative to the description of its object. There is no desire for anything that is not conceived in some way. But to conceive is not necessarily to believe. It is the conception of an object that is a necessary condition of desire, in the sense that desire is always relative to a way of conceiving an object. It is not necessary that this conception amount to belief. The person who desires (momentarily, we hope) the destruction of the whole world (in preference to the scratching of his finger) must imaginatively conceive of that state of affairs. His beliefs are not directly implicated in this desire.

In Hume's example of the fruit that is believed to be sweet, it is difficult to see this potential split between desires founded on belief and desires that are parasitic upon mere conception. This is partly because there is little riding on the adjustment that takes place when the "excellent relish" I have imagined is replaced by the belief that it is sour. But change the example. Concerned for my partner's welfare on a stormy evening, I have wanted to speak with her. But now the dreaded policeman at my door has convinced me that this is impossible. Does my desire to speak with her vanish because I have been persuaded of this? Of course it does not. It is almost inevitable that I will continue to imagine our world as is was before the accident. Our normal reactions would be incomprehensible if our desires did track reasonable beliefs in this situation.

(2) Grief: But perhaps there is an alternative explanation of grief; one that does not involve wanting what one believes to be impossible.¹⁴ When anyone close has recently died, the required changes in one's beliefs and dispositions are profound and it takes time for these changes to occur. One finds oneself reaching for the telephone, then realizing that one can no longer call. Perhaps, in the emotional confusion that surrounds the death of a loved one, one actually forgets—the belief that the person is dead disappears. One finds oneself wanting and even preparing to call because one has slipped back into a system of beliefs that no longer applies. This really is a case of the sort that Hume describes in commenting on the sense in which our passions can be unreasonable: the desire to phone is founded on a state of affairs that does not in fact exist. When one recovers from this momentary lapse, then (normally) the desire to phone does "yield," just as Hume says.

We do not want to deny that this accurately depicts one of the ways that we can be affected by a person's death. A full discussion of this phenomenon would have to confront the fact that, like desires, beliefs are not simply deliverances of reason. But we do deny that this kind of cognitive dissonance provides an alternative to an account in which one continues to want what one believes to be unobtainable. One returns the phone to the receiver feeling absurd, perhaps. But the longing for communication with the person one has lost will not have disappeared. Nor would one experience grief at all if desires of this sort disappeared with the realization that their fulfilment is impossible.

7. Humean Explanations

Why does Hume think that passions must "yield to reason" with the discovery that they cannot be satisfied? On the face of it he has a surprisingly strong argument— one that we have already seen. We found that, on Hume's account, having a desire for something requires a belief that there is a prospect of pleasure from that object. As we have seen, Hume says "'Tis obvious, that when we have a prospect of pain or pleasure from any object, we feel a consequent emotion of aversion or propensity" (T 2.3.3.3; SBN 414). Consider this claim in relation to the passion of grief. The belief that a person exists is certainly an element in the causal conditions that bring about the desires that constitute caring about someone. It also seems safe to suppose that when the cause of something is removed, the effect must disappear. So, on this set of assumptions, it is natural to think that change of belief (that this person exists) will remove one of the causal conditions of the desires, and hence that the desires themselves will vanish.

Clearly, for some causal relationships this type of inference is correct. If I remove the object (cause) which is casting a shadow across my page the shadow (effect) will disappear. But many causal conditions are not of this sort. The relation between parent and child (one of Hume's own examples) is also a causal relation. Obviously, the parent's going out of existence does not necessitate the demise of the child. Which of these paradigms seem best to fit the relationship between the (believed) loss of a spouse or friend and the system of desires that constitute caring for the person?

Our previous discussion offers our answer to this question. These desires often "continue in existence." They are not like a shadow that must disappear with the removal of its source; they are like a child, whose existence is maintained by causal conditions that are separate from the causes that brought him or her into existence. The Humean mind is Newtonian. Like bodies set in motion, perceptions (and this for Hume includes passion) have properties analogous to momentum and inertia. As we have suggested, Hume's principles of association are central to understanding the properties of inertia or momentum that Hume relies upon in his understanding of the persistence of passions in the absence of their source. This would certainly be part of the Humean explanation as to why our passions do not always yield to reason's discovery that fulfilment is impossible.