We are unable to display your institutional affiliation without JavaScript turned on.
Browse Book and Journal Content on Project MUSE

Download PDF

A Conversation between Annette Baier and Anik Waldow about Hume's Account of Sympathy
Annette C. Baier (bio)
Anik Waldow
Anik Waldow is lecturer in the Department of Philosophy at the University of Sydney. E-mail: anik.waldow@arts.usyd.edu.au.

We discuss the variety of sorts of sympathy Hume recognizes, the extent to which he thinks our sympathy with others' feelings depends on inferences from the other's expression, and from her perceived situation, and consider also whether he later changed his views about the nature and role of sympathy, in particular its role in morals.

Annette Baier: Hume wrote in 1751 to his friend Gilbert Elliot, who had been trying to come to grips with the Treatise account of causation, that he should stick with the first Enquiry, and not bother with the Treatise, which he said he had much regretted allowing into print, without having revised it. Presumably one thing he wished he had revised concerns causation. Then in 1775 he instructed his publisher, William Strahan, to have inserted into all subsequent editions of his works that it was to his post-Treatise works that readers should turn to find his "philosophical sentiments and principles," since in them he had corrected "some negligences in his former reasoning, and more in the expression." Are any of those negligences to be found in his Treatise account of sympathy? His "anatomy" of it does involve several causal claims, but surely he does want to keep the claim that we are capable of some sort of sympathy, however exactly that is best analyzed, and his version of morals seems to require that we exercise it, as a preliminary to moral judgment, [End Page 61] where this is taken to be a special pleasure or displeasure in character traits. Admittedly, Enquiry concerning the Principles of Morals (hereafter "EPM") gives a slightly different version from the Treatise, both of the point of view from which recognition of virtues and vices is made, and of the role of sympathy,1 but he still says that morality shows "the force of many sympathies" (EPM 9.11; SBN 276).2

Anik Waldow: In the Treatise, Hume first describes sympathy as a process that requires us to form ideas of feelings before we are able to develop feelings with other persons (T; SBN 317)3 ; and even in EPM he emphasizes that our imagination needs to be stimulated before we can emotionally engage with one another (EPM 5.37; SBN 224). The question that arises here is why Hume puts so much emphasis on the need to form ideas of other persons' passions and circumstances before being able to engage emotionally with them. My suggestion is that he wanted to tell us that ideas and beliefs about other persons, their mental states and situations, are important to our ability to respond emotionally to each other and to have moral sentiments.

Hume writes: "The approbation of moral qualities most certainly is not deriv'd from reason, or any comparison of ideas; but proceeds entirely from a moral taste, and from certain sentiments of pleasure or disgust, which arise upon the contemplation and view of particular qualities" (T; SBN 581; my emphasis). Hence, he tells us that moral sentiments arise from certain "views" and "contemplations." Furthermore, consider that the "pleasure" and "disgust" that Hume here takes us to feel in another person's character need not be the feelings the person herself has. He does say we must expect others to agree, when we feel approbation, but whether we must all feel the same pleasure, or sympathize in the sense of feel with each other's moral pleasure, is not so clear. How in any case could we ever know if our pleasure in good qualities, and disgust at vices, feels just like another's?

Once the need for views and contemplations is appreciated, one needs to specify what these views and contemplations are, on which the moral sentiment depends. It seems to me that Hume here alludes to ideas rather than feelings, for he speaks about our thoughts rather than our emotions about other persons and their qualities. If so, that would suggest that moral approbation can proceed via ideas, because it is ideas which put us in a position to develop moral sentiments and allow us to approve of someone as a virtuous or vicious person. These ideas concern the character traits of the person who is to be judged and, when we adopt a general point of view, the feelings of those persons with whom the character in question has commerce.

If we pay attention to the ideas involved in processes of sympathy there are two advantages. First, it becomes comprehensible how we can come to a "steady and general point of view" (T; SBN 581–82). According to Hume we adopt this point of view by transcending our own perspective. That is, we consider not [End Page 62] only the feelings another person's conduct causes in us, but the influence she has "upon those who have an intercourse with any person" (T; SBN 582). It is clear that the greater the range of persons affected by the character in question, the more difficult this task becomes—at least if it is required that sympathy must result in shared feelings, for it is rather challenging to have the emotions of let us say fifty people at once. There is no such difficulty, however, if it is assumed that sympathy need not result in empathic feelings but can result in ideas about another person's feelings, thoughts and beliefs. It then looks as if we can judge a character by means of a sort of sympathy, even if this means that we have to consider the emotional responses of a great variety of people. We can do this by forming ideas of these emotional responses, ideas that can represent ideas of similar emotions, as Hume tells us when he describes the associative process by which ideas "are not really and in fact present to the mind, but only in power" (T; SBN 20). Extensive sympathy with all parties here becomes an easy task: all we need to do is form ideas that represent the possibly vast variety of passions other persons experience when dealing with the character that is to be judged.

To conceive of sympathy as a process that enables us to have moral sentiments by the formation of ideas and beliefs about other persons, their emotions and situations, rather than by merely having empathic feelings, is to be recommended for another reason. As will emerge in due course, this understanding of sympathy allows us to see that sympathy proceeds upon associations that the perception of ourselves establishes. Why this is so becomes clear if we take a closer look at the conditions that need to hold if ideas of passions are to appear. Hume is clear that "the ideas of the affections of others are converted into the very impressions they represent, and that the passions arise in conformity to the images we form of them" (T; SBN 319). This suggests that our ability to feel someone's passion first and foremost requires our ability to conceive of the passion in question, because it is an idea of the passion that the mechanism of sympathy converts into the passion itself. Furthermore, Hume takes passions to be "simple and uniform impressions" (T; SBN 277).4 This entails that we can form ideas only of those passions that we have once experienced ourselves, for Hume claims that "all our simple ideas in their first appearance are deriv'd from simple impressions" (T; SBN 4). Of course, this is not to deny that these ideas can represent passions which we have never felt ourselves: to understand that you are in pain and to feel for you does not require me to be familiar with the precise pain you are enduring.5 All that is needed for my sympathy with you is that I have once felt some sort of pain myself, for then there is an idea available by which I can represent and understand your pain. I here sympathize with you without paying attention to the particular back-pain from which you are actually suffering, back-pain that I have never experienced myself and of which I therefore cannot form an accurate idea. More generally speaking, one can say that the ideas that are required for our emotional responses to other [End Page 63] persons are the after-images of emotions that we have once felt ourselves, after-images, however, that enable us to understand emotions that we sometimes do not know from our own experience, in the exact form another has them.

According to Hume, these after-images are "conceiv'd to belong to another person" (T; SBN 319) when we immediately pass from "the effects of passions in the voice and gesture of any person … to their causes" (T; SBN 576), causes that are taken to be the passions that we end up feeling if our original conception of them becomes lively enough. Hume here explains what we do when confronted with another person, but he does not give us any details about why we pass from impressions of another person's conduct to the idea of a passion as the cause of the perceived conduct, a passion that we have once experienced ourselves.

To answer this question I think that we need to take a closer look at Hume's causal language, when he says that we pass from the expressive voice of gesture of another to their causes, in the passions of the other. According to Hume's definition, something qualifies as a cause if it appears in a constant conjunction to something else and if the perception of this thing (or event) determines the mind to form an idea of the thing (or event) that usually accompanies it (T; SBN 171). From this definition it follows that passions can be considered as the causes of behavioral expressions only if they are perceived constantly to precede behavioral effects and thereby determine the mind to link ideas of a passion to impressions or ideas of behavior. Passions of other persons are imperceptible. Hence it must be self-perception, that is, the perception of the constant conjunction between our passions and our behavioral expressions and the resemblance between the other's behavioral expressions and our own that makes us associate after-images of our own passions in reaction to impressions of another person's conduct. Note that we associate in this way, that is, on the model of our own past experience, only if we do not consider further sources of evidence, such as the reports and stories about other persons' emotions or situational contexts. Information of this kind can trigger different associations, or even directly invoke an impression of a passion, namely when we are emotionally affected by the reports and stories of another person's fate. Our own experience is where we begin, even if we go beyond it.

Cases in which we engage in sympathy-directed inferences and those in which we infer from cause to effect, or effect to cause, in the physical world are both built on the habit of perceiving constant conjunctions, and in this respect can be seen as being structurally analogous. However, it needs to be noted that the first stage in processes of sympathy requires us to transform the very experience that originally enables us to associate ideas of mental states with impressions of behavior. That is, we join ideas of first-person experiences of passions to impressions of third-person behavior, despite the fact that the kind of experience by which we come to associate in the alleged way results from the perception of constant conjunctions between the first-person experience of certain classes of passions and the first-person [End Page 64] experience of certain classes of behavior. Now, this change of perspective is certainly foreign to ordinary causal inferences about physical objects, inferences which are not concerned with the experiences of other persons, such as their thoughts, sensations, and passions. In inferring that a given physical object will react in a particular way, we keep precisely the perspective inbuilt in the experience that informs our expectations: we consider classes of resembling objects and their previously observed reactions from a third-person point of view. This suggests that sympathy is a particular kind of inference: it shares with causal reasoning its associative and inferential structure, but it involves a shift in perspective that we do not encounter in most ordinary cases of inference, which are concerned with phenomena in our perceptible surroundings.6

AB: I think Hume's science of man has plenty of cross-perspective inferences, that they are not peculiar to sympathy. He infers that the child will get the same idea of orange color he does, when given an orange, just as the one who has not tasted pineapple, but knows those who have enjoyed its taste, expects to enjoy it too, when he gets to taste it. At least before part 4 of Book 1 of the Treatise, Hume did suppose that we all share one world, hear the same sounds as those around us, speak a shared language, read some of the same books, enjoy much the same tastes, and so on. The perceptions of others may be imperceptible to us, but they are regularly taken to occur. The very first paragraph of the Treatise, gives, as an example of impressions, what the reader sees on the page as she reads, and of ideas, what meaning she gives those words. And as for the causal link between emotions and their typical expression, I think that, though we may see our own gestures and hear our own tone of voice, for many other involuntary expressions, especially facial ones, we do not usually observe them. We are likely to come to believe that frowns express displeasure as much by seeing others frown and asking them why they are creasing their brow like that, and then believing their answers, as by noting any correlation between our own displeasure and our frowns. We do know the latter from the inside, by proprioperception (or maybe by our mirror neurons), but, unless we are Narcissus, we do not know our expressions as others see or hear them. We may be taught what smiles and frowns mean, from picture books, or stories. Hume claimed that "more than one half" of our beliefs, presumably including causal beliefs and beliefs about expression, are said to be due to education (T; SBN 117). Although this was said in the Treatise, it is true that in other places in it he put stress on what our own experience teaches, so the question of how we combine our own experience with our experience of others, and what they tell us about their experience, is never directly addressed. There is some discussion of testimony, but that is all we get on knowledge-sharing.

As for the special case of cross-perspective switch which occurs in sympathy, I think we need to see it as more than just thinking of a familiar feeling, when [End Page 65] observing another's expressive behavior, behavior which is believed to be the effect of such a feeling, and attributing that feeling to the other. For the torturer could think of the pain he is causing his victim, without showing any sympathy. The first stage of Hume's sympathy process could lead to malice, not sympathy, and on his view it sometimes does, when we hate the person whose suffering we observe, or are too preoccupied with our own situation to bother with any sympathy. So for sympathy, in addition to inferring what the other is feeling, there must also be some understanding of what the other feels, and an appreciation of his wish, if it is an unpleasant feeling, that it cease. Even when malice occurs, there may briefly be some sympathetic sharing of the other's distress, before malicious pleasure drowns it out. Even when "the principle of comparison" operates, and we treat the other's feeling primarily as a standard of comparison with our own, Hume speaks as if the displeasure we then take in a competitor's success comes after a brief sympathetic sharing of the pleasure.7 "The direct survey of another's pleasure naturally gives us pleasure, and therefore produces pain when compar'd with our own." (T; SBN 376). In EPM 6.33n34 (SBN 248n), he says that even when we feel envy, it either coexists with or alternates with respect. We first feel pleasure with the other, then pain when we compare his success with our own relative failure. Sympathetic sharing of the other's feeling is the natural response to knowing of it—that is, the first step in the sympathy process is naturally followed by the second—when we are in the other's presence and so have a "direct survey" of his expression of his feelings, unless we are so preoccupied with our own state that the principle of comparison kicks in. That Hume can speak of direct survey of another's pleasure, not just of its expression, is interesting, and suggests that the cases where we read or hear of another's feelings and situation are "indirect," in comparison with those cases where we are with the person and see her easily-read face, or hear the anger in her voice.

Somehow we do come to have beliefs about how we all express our feelings, and about the extent to which we do all usually understand each others' expressions and feel sympathy, and many sources, including our education, may contribute to that. There may be a more explicit acknowledgement of knowledge-sharing in Enquiry concerning Human Understanding (hereafter "EHU"), especially when we accept "inviolable laws of nature," presumably because we believe what scientists tell us. In his long footnote to the section "Of the reason of animals" on why some reason better than others about causes, he cites not only the "greater experience" some have, but also the "enlargement" of experience some receive more than others do from "books and conversation" (EHU 9.5n20; SBN 107n).8 I have argued elsewhere9 that Hume made considerable changes in EHU, after the reviewers of the Treatise had accused it, among other things, of "egotisms." The long footnote on the causes of variation in the causal reasoning of men and animals is one such change. [End Page 66]

This change was one of enlargement on his treatment of the topic of the enlargement of experience, not of correction of an outright error. But an important thing that Hume did correct in EHU was his double definition of cause. For if we stick by the Treatise definitions, requiring spatial contiguity between cause and effect, and some impression of the cause, or of the effect, from which we infer its causal partner, then no perception of our own, what you term a first person experience, could count as what we take to be a cause, or an effect, since perceptions lack spatial position (or so Hume says) so cannot be contiguous to one another, and we have no impressions of them. Some of them are impressions, but we do not have meta-impressions of them. Far from being confined to his own experience, in the Treatise, Hume has official trouble in letting its findings even contribute to his causal beliefs, though I do not think he realized he had a problem until the Appendix, at the earliest. The EHU definitions of cause mention neither impressions nor contiguity. Instead of an impression of the cause, giving rise to an inference to the effect, he speaks of the cause's "appearance" to the mind. Since we are conscious of our own perceptions, they do appear to our minds when they occur, so perception-causes, in our own case, are let in by EHU's definitions, as they strictly were not before. Hume takes his own mind to be a typical human mind, so uses it as his specimen, when talking about perceptions. In contemporary terms, he uses his own to "simulate"10 those that others are taken to have. Almost all the main theoretical claims in his Book 1 account of "the understanding" in the Treatise, the claim that ideas are derived from impressions, the claims about association, belief formation, volition-causes, and the Book 2 claims about sympathy, were claims about causal dependencies between perceptions within one mind, but they were not covered by the official definitions. This was an embarrassing incoherence in a work that had claimed greater "coherence" in its parts than previous systems. The fault was in the way the definitions were expressed, more than in the substance of the theory. So once the definitions were revised, one would expect most of the causal claims about our minds to survive in later works. But many, such as the long account of how we form probability estimates, and the elaborate account of how sympathy comes to occur, are missing in the later works.

Sympathy is referred to briefly in part 2, section 10 of the Dissertation on the Passions,11 and plays a role in EPM, but the elaborate account of how it occurs in us, how we get ideas of others' joys and sorrows, which may be converted into empathic impressions, so that we come to feel with others, is not given in either place. In EPM 5.17n19 (SBN 219–20n), Hume says it is needless to ask why we have fellow feeling with others, and in EPM 6.33n46 (SBN 248n), Hume says the workings of sympathy are "extraordinary and unaccountable." There is a reference to the "enlivening of sympathy" by eloquence in EPM 5.43 (SBN 230), but in the Treatise nothing seemed to count as sympathy with someone else's passion if it did not, or at least if it might not, end in a lively impression, in fellow-feeling. [End Page 67] Sympathy in the Treatise can be more or less "extensive" (both in the Book 2 sense of extending over a fair stretch of a person's life, and in the Book 3 sense, of extending to many people12 ), and more or less "deep" in its understanding of the other's reasons for feeling as she does; it can be stronger or weaker, and can be drowned out when "the principle of comparison" makes us pained at another's pleasure, when we envy it, or pleased at their pain, when they are our enemies or rivals. But if it were just a vivid idea of how another is feeling, or of the predicament she is in, the sort of thing eloquence could produce, that would not count as real, completed or full sympathy, unless our own feelings "reverberated" or echoed those we are told about. "In sympathy there is an evident conversion of an idea into an impression" (T; SBN 320). "The minds of men are mirrors to one another, not only because they reflect each other's emotions, but also because those rays of passions, sentiments and opinions may often be reverberated" (T; SBN 365). Hume also says we sympathize with each other's sympathy for us, so the transfer of sentiments by sympathy need not be one-sided, and is capable of several "rebounds." He says the chief "recommendation" of riches is that they produce esteem in others, who sympathize with the possessor's satisfaction in them, and the possessor then sympathizes with those who esteem him, a "third rebound" of the original pleasure. This talk of reverberated sentiments is talk about our feelings, not just our ideas. Certainly animal sympathy, as Hume discusses it, seems to be communicated feelings, not just animal awareness of how other animals are feeling, and Hume does say there can be cross-species sympathy. Even if, as you plausibly maintain, only ideas of others' feelings may be needed for the sort of wide-ranging sympathy, and sympathy with the public interest, that the moral sentiments require as preparation for their operation, the sort of sympathy with individuals that Book 2 spoke of did seem to be empathic feeling.

AW: The main focus of Book 2 of the Treatise is laid on the description of sympathy as a mechanism that enables us to share feelings, either after inference to them, or more spontaneously. Hume here emphasizes that sympathy sometimes takes the form of contagious emotions, for instance when he points out that we feel the pleasure of the rich in virtue of our sympathy with their satisfaction, but not by forming ideas about the beauty or usefulness of their properties. And you may be right in claiming that animals do sympathize by feelings rather than merely ideas, although in principle they can sympathize by ideas, if sympathetic feeling presupposes ideas of another's emotions. However, I do think that Hume's conception of sympathy, even in Book 2, is not exhausted by those cases in which sympathy leads straightforwardly to emotions. What leads me to think this is that in Of the Love of Fame (T 2.1.11; SBN 316–24) Hume spends a lot of time describing how much our relations with each other matter with respect to the intensity of our sympathetic responses to other persons. This seems to suggest that it is not necessary for us to [End Page 68] feel with every single human being, in order to have "the sentiments of humanity and sympathy" (EPM 9.12; SBN 276). There is more room for variation in vivacity in Humean ideas than in impressions, which must all reach a certain threshold of vivacity to count as impressions, so the range of strength which sympathy shows needs ideas as well as impressions in order to show itself.

Of course, the idea or impression of the self is supposedly always present to us and in principle ready to turn every idea of another's emotion into the emotion itself. Strictly speaking it is, however, not the perception of the self that is attended to, but the "relation of objects to ourself" (T; SBN 320). Indeed if we are too self-preoccupied, sympathy may be blocked.13 In order to enliven an idea of another's passion to the extent that it becomes the passion itself, the right relations are needed, namely relations that connect the idea of self with the ideas of other persons, and connect us to them not as rival or enemy. These sympathy- promoting relations can be stronger or weaker, as Hume explains when emphasizing that we "sympathize more with persons contiguous than with persons remote from us: With our acquaintance, than with strangers: With our countryman than with foreigners" (T; SBN 581). Here then it becomes conceivable that in some particular cases, contrary to what Hume says in other places,14 the "great resemblance among all human creatures" (T; SBN 318) is insufficient for causing sympathetic emotions. In these cases the relation between the idea of the self and the idea of another person would be too weak to allow the idea of another's passion to acquire enough liveliness to qualify as an empathic feeling rather than a mere idea or lively conception of another person's emotion.

Hume furthermore writes that, "however instantaneous this change of the idea into an impression may be, it proceeds from certain views and reflections, which will not escape the strict scrutiny of a philosopher, tho' they may the person himself, who makes them" (T; SBN 317; my emphasis). Presumably, these views and reflections relate to the enlivening of the ideas of another's emotions, and hence to the relations that connect us with other persons "beside the general resemblance of our natures" (T; SBN 318; my emphasis), relations that are founded on "any peculiar similarity in our manners, or character, or country, or language" (T; SBN 318). Once again this suggests that there is a point at which mere species resemblance between human beings is insufficient for causing sympathy qua empathy, that is, sympathy qua shared emotions, that emerges only in a second step. If this point is reached we need to re-view our relations to the other and reflect on her particular situation and thereby support the otherwise natural vivacity transfer in order to feel with the other.

Repetition of ideas may be another way to enliven our conception of another person's emotion. Thus Hume emphasizes that "we may feel sickness and pain from the mere force of imagination," however, not before "often thinking of it" (T; SBN 319). Hume furthermore admits that the difference between the [End Page 69] idea of someone's pain and the impression of her pain "may be remov'd, in some measure" (T; SBN 319; my emphasis). He thereby gestures towards the possibility that in certain situations we plainly do not take the alleged measure and only form ideas about another's feelings rather than the feelings themselves. And this supports the view that sympathy qua empathy is not essential for some understanding of what another is feeling.

The inference to the sensation a child encounters when seeing an orange certainly requires us to take measures of a different sort. Here we are not interested in reviving a sensation we have once felt ourselves. We merely reach for an understanding of the other's sensation and can do this by associating ideas that situations of a similar kind usually trigger in ourselves. We thus engage in the above-mentioned change of first- to third-person perspective that we need not perform if we reason about the causes and effects of physical objects. So if we infer to another person's mental states, either explicitly or implicitly by spontaneously associating ideas upon our own past experiences, we perform precisely what we perform when engaging in the first step of sympathy, only that sympathy, at least in its first stage, is always spontaneous and, as you have pointed out, leads further, for it requires some sort of seconding of and concern for the other.

AB: Sympathy is special in our observation of one another because its second step makes us even more like the other than we already were. Hume, like Malebranche, thinks it, along with imitation, helps explain national character. You began by putting the emphasis on the first step in the full sympathy process, that resulting in a belief about how another is feeling, and perhaps also about why she feels that way, or what convictions she holds, and what leads her to hold them, since this is where Hume starts, in Book 2, part 1 of the Treatise, when he first discusses sympathy. And this is what is needed for morality, as long as it includes some understanding of why the other feels as she does. You also stress Hume's claim that, though the second step, to feeling with the other, when it occurs, may appear instantaneous to the sympathizer, the philosopher can see it to depend on some reflection. I would add that he also believes the first spontaneous step can, not only appear to be instantaneous, but actually be so. Like sub-atomic particles, we can be "entangled" with each other. As I have pointed out, he says that we can have a "direct survey" of another's pleasure and he also says we can have what he, following Shaftesbury,15 calls a "pre-sensation" of what another must be feeling (T; SBN 322). (Hume spells the word without a hyphen.) Do you take this to be a sensation, or just an idea of what the other is feeling, an advance warning of what we may come to feel, in sympathy? In cases where we see the other's easily-read expressive face, and recognize what it expresses, the analogy to our own case need not be spelled out. And it is in any case implausible to suppose that we always have noticed the way we ourselves express [End Page 70] our feelings, as if, when babies, we had mirrors to see our own smiles. We know our own smiles and frowns from the inside, as it were, not by observing them, and they are usually involuntary, so not a case of what Hume takes the will to be, a change in movement of the body that we knowingly bring about. We may have "knowledge without observation"16 of our smiles, as well as of what we are intentionally doing, but not till others pointed it out to me did I realize that I express doubt by tilting my head to the right. And even when we do know what expressive gesture our body is making, we do not, in Robert Burns' words, have the gift to "see oursels as ithers see us."17 Recent discoveries of mirror neurons explain how we have this quick recognition of what others' bodily expressions indicate about their immediate intentions and feelings, and autistic people are handicapped by lacking it. They must rely on the argument from analogy, which of course can occur, and which Hume thinks often does occur, especially in cases where it is the circumstances the person is in, not his expressive behavior, that prompts our sympathy. Then we must ask ourselves how we would feel, in their situation, for example if about to have surgery without any anesthetic, or trampled by galloping horses. Surely you agree that he thinks we are "like strings equally wound up," so do react similarly to the same stimulus, and also that we can have quick recognition of what another feels when we see her expressive face, or hear her anguished voice, but, in cases when, say, her face is not turned in our direction, and she keeps stoically silent, we can infer what she will be feeling, from our knowledge of what we would feel, in her circumstances? (We could call this Smithian sympathy, since he takes all sympathy to be of this sort. Even when we see the downcast face of another, he says, we ask "What has befallen you?" and wait for an answer, before feeling sympathy.18 )

AW: Hume, as if anticipating simulation theory, writes that pre-sensations tell us "what will operate on others, by what we feel immediately in ourselves" (T; SBN 332). He raises this point when arguing that humans react to the same sort of "qualities and circumstances" (T; SBN 332) with the same sort of passions, so, in general the causes of pride will also be causes of love. I take this to indicate that pre-sensations are impressions that we share with others when being affected by the same sort of situation as those persons with whom we share the feelings we take them to have. It here seems that pre-sensations may indeed do the work that modern science attributes to mirror neurons. So I grant your point: Hume allows for an understanding of passions that does not involve ideas of passions that derive from our own past experience. But I would like to add that pre-sensations require the sympathizer and the person with whom she sympathizes to be affected by the same situational context, be this by being directly present at the situation that causes the other to feel the way she does, or vividly recreating it after reading or hearing about it. In this sense I would say that pre-sensations just are experiences, [End Page 71] namely experiences that we have when being emotionally affected by a situation, report or story about another person's fate.

If we think of pre-sensations as the direct experience of emotions that emerge in situations of a certain type, pre-sensations indeed turn out to enable us to feel what you called "Smithian sympathy": they arise upon the perception of a scene which may or may not include the other's facial expressions, such as the "anxiety and concern" in the eyes of the "patient and assistants" before one of "the more terrible operations of surgery" (T; SBN 576). This does not mean that Hume takes it that we always sympathize by having pre-sensations. Hume first introduces sympathy in T 2.1.11 (SBN 316–24) by describing a process that begins by the formation of ideas, before these ideas are turned into impressions. This shows that we need to distinguish (at least) two forms of sympathy: that which proceeds by pre-sensation impressions, what we could call the pure contagion cases, and that which first takes us to the formation of ideas, which can then be converted into impressions, what we could call the cases of a special kind of inference from analogy.

When I say that sympathy is grounded in the habit of perceiving ourselves and the associations of causation and resemblance, I refer to the latter kind, the kind of sympathy that refers us back to our past experiences by leading us to ideas of passions, ideas that, as long as we grant passions to be simple impressions, we can have only if we have once experienced the passions ourselves. As I said before, the origin of these ideas of passions, however, does not prevent them from representing things that we have not experienced ourselves. So when Adam Smith says that it is possible for men to sympathize with a woman's pain when giving birth, Hume could agree with this.19 He only would need to accept that men sympathize not by forming an exactly similar idea of the pain in question, for men just lack the experience of what it feels like to be in what Smith calls child-bed, but by entertaining any idea of pain which is a copy of one of their own previous pains and can represent the pain of the woman.

Note that although it is a causal process that underlies idea-based processes of sympathy, it would be false to claim that we first need to consult ourselves and choose one particular idea of a previous experience as the most probable candidate for explaining the other's behavior. The behavior of the other person, and our perception of it, prompts in us an idea of a certain passion that we immediately ascribe as the cause of the perceived behavior. In this sense, impressions of third-person behavior function as triggers that set off a natural associative process which results in the conception and belief in another's mental state.

So of course I would agree that Hume thinks that human minds are mirrors to each other, are equally "wound up," wound up both to respond in the same way to the same situations, and to respond to one another, and that sympathy indeed enables us to understand and share the emotions of other persons. But it needs to be remembered that it is experience, be this the past experiences of our [End Page 72] own passions, or the actual experience of pre-sensations, or hearing reports of others' passions, which provide us with ideas of passions by which we can engage in sympathy.

AB: I agree that the pre-sensations include the cases where we put ourselves in the place of another, when we wince when another is about to be struck, a favorite example of Smith's. But I doubt that men have mirror neurons telling them what childbirth feels like, so men's sympathy with women in labor will, unless they suffer from Couvade syndrome, and so imitate the whole process of pregnancy, from morning sickness and weight gain to final labor, will not be by pre-sensation. And I think the pre-sensations, if they really do anticipate mirror neurons, also include those "contagion" cases when we see the face of another, have a "direct survey" of their pleasure or displeasure, and know immediately what that one must be feeling, partly because we in some sense know, from experience and maybe from our mirror neurons, how the face muscles must move to produce that expression, and know what feeling triggers that movement. So it is not only Smithian sympathy that can be known without any explicit associating of ideas of passions with sense perceptions of situations and expressions. I suppose we could say that we put ourselves behind the face of the one who is frowning, or smiling, so turn this case into Smithian sympathy. But why turn Humean sympathy into Smithian? Smith is in any case generally rather dismissive about sympathy with pain, saying that pain is soon over, and we do not sympathize even with our own past agony. He thinks sympathy is usually for some state involving the imagination as well as bodily feelings. "Pain never calls forth any lively sympathy unless it is accompanied with danger."20 But Hume does think we usually sympathize with the one in obvious pain, as well as with the one suffering no pain, or even fear, but in great danger, like the infant prince taken captive by enemies. There is room for many sorts of sympathy in Hume's account, including the cases he began with, where we do begin with an idea of what another is feeling, rather than with any feeling of our own, and it is those cases which are important for his account of how the moral sentiment depends on sympathy.

As for the bases of our causal inferences and causal associations, our own experience will be the natural place to start, but usually we will "enlarge it," accept testimony about others' experiences, imitate others, and let the great tragedies teach us about emotions and their expression. Hume does retreat to a solitary stance in Treatise Book 1, part 4, but before that he had cheerfully generalized about other people's experiences, taking it that there was a common human nature to investigate, that there were gender stereotypes, and national stereotypes (witty Frenchmen, witless Irish). Even the reason and passions of animals came within his purview. He knows what makes dogs howl as well as what makes human beings howl. In his discussion of how we form probability estimates, he contrasts [End Page 73] the peasant's with the artizan's reaction to the clock stopping, and he discusses how we accept what historians tell us about the death of Caesar. So quite a bit of knowledge-sharing went on, in Book 1. Only for explaining our idea of necessity is it really necessary that the constancies we rely on, in causal inferences, be in our own experience, and even there it is the inferences themselves, even if they rely on what we have been taught, that are needed to explain the idea of causal necessity. But I think Hume is more explicit in EHU than in the Treatise about how much our causal inferences do owe to the pooling of information. So maybe he saw his stress on individual experience, in his Treatise account of sympathy, a stress you are making stronger, as a defect. It was one of the aspects of the Treatise picked on by those who opposed his candidature for the Edinburgh chair. These critics, to whom he replied in 1745 in his "Letter from a gentleman to his friend in Edinburgh," had singled out his claim that, even when we attend to things outside ourselves, "[w]e never really advance a step beyond ourselves" (T; SBN 67), and he had already been accused by his first reviewer of "egotisms."

At any rate there are, after the Treatise, no pre-sensations, nor any inferences from the presumed analogy of other's expression of feelings with our own to conclusions about their feelings, nor any explicit cases of our coming to literally feel with others. I offer four suggestions as to why. First, by the "Dissertation on the Passions," the term "impression" is phased out. In EPM the sharp contrast between impressions and ideas is not drawn, and the term "impression" is used in a more colloquial sense for any influential impact on the mind (I have elsewhere discussed these changes,21 which may be no more than terminological.) So, by EPM, Hume would no longer describe the second step in the sympathy mechanism as he did in the Treatise, as a conversion of an idea into an impression matching it. Second, the account of the process of sympathy, in the Treatise, is a case of a causal story about how perceptions with some determinate content and vivacity cause other perceptions with related content and derivative vivacity, and no such causal stories about perceptions causing other perceptions were covered by the Treatise definitions of cause. The second definition required that we have an impression of either cause or effect, presumably as guarantee of its really happening, and we have no impressions of perceptions. Perhaps the closest we get to impressions of perceptions is when we see another person seeing or undergoing something, and perhaps feel sympathy. Suppose we are nurses who see someone enter a hospital room and find there a loved one who has just been badly injured. We see their seeing, or at least see their looking, see their expression of distress, and may feel with them. That would be having impressions of others' impressions, or at least of the bodily indications of them. But in our own case, we know we are having an impression by immediate consciousness, not by seeing ourselves seeing, or observing our own distress. And it is our own case that Hume seems generally to be appealing to in his theory of the mind, in the elaborate causal stories he gives about how our inferences occur, [End Page 74] how our beliefs are formed, how our volitions depend on our desires and beliefs, how we come to share the joys and sorrows of others. He assumes each of us does take herself as a typical person, and that it is natural to at least start with our own experience. The account of the mechanism of sympathy is as much implicated in the incoherence between his official definitions of cause and his own causal claims about our nature, as all the other cases where either cause or effect is mental, so is not located, and not the object of any impression. The only time Hume relies on what we know of others, to correct our mistaken views of ourselves, is when he points out how predictable we find others' decisions, while we like to think our own not so predictable. Usually he takes us to have immediate access to our own mental states, and to be able to infer effects from causes there as readily as we do elsewhere. Do we need to have "post-sensations" of our own impressions, to know they have occurred, and to expect them to have causal consequences? Surely not. Consciousness ensures that we are aware of our own impressions.

I believe that this major incoherence in the Treatise is one main reason why Hume did not want us to study it, but to turn our attention rather to his later writings. The definitions of cause are revised in EHU, no longer mentioning either contiguity or impressions. Perception causes and effects are made room for by the revised definitions. So why did he not repeat the earlier accounts of the sympathy mechanism? To explain that we must suppose something else was wrong, unless we think he was simply trying to be brief. Another Treatise account that is shortened and changed in the later writings is the general account of how we form our beliefs and probability estimates about matters of fact. The first stage of sympathy is a special case of belief formation, when we form a belief about another's state of mind from some sense impression of another's expressive behavior, or by imagining, from their circumstances, what they must be feeling. Both in the general account of belief formation and in its application to how beliefs about another's mental state sometimes lead to sharing that state, the Treatise had claimed that an idea gets upped in vivacity, because of the ready availability of some current related impression, which can share its vivacity. In the first stage of sympathy an idea of the other's feeling is upped into a belief, by its causal relation to the feeling's observed expression. In the case of the second stage of sympathy, this source of vivacity, to convert a belief about a feeling into an instance of that feeling, was "the idea or rather impression of ourselves." (Of course if this idea or impression is too dominant, it blocks rather than enables sympathy.) Hume knew his Book 1 account of our idea of ourselves as identical over time was defective: he said so in the Appendix. And he had in Book 1 denied we had any impression of ourselves as minds. So there was a problem. (It may not be an insuperable one, if we suppose that the idea of self involved in sympathy is not of a bare mind, but of an embodied, passionate, expressive and social self, one with the relations to others which, as you have stressed, determine how strong [End Page 75] our sympathy with them will be. But how an idea of self could both enable and block sympathy was, I think, a more intractable problem.22 ) He did not address the topic of our self-conceptions at all in the Enquiries and Dissertations, except to discuss our views on whether our wills display causal power, or freedom. So any problem there may be with our "idea or rather impression of ourselves" is a third reason why the detailed account of how full sympathy occurs is not repeated. But this was involved only in the second step of the full sympathy process. He had not relied on any impression of self in the general story about how ideas become beliefs, where any current impression with some suitable relation to the belief would do as source of the belief's vivacity, that would not explain why the account of belief-formation in EHU is so truncated, and why instinct, not vivacity-transfer from an impression, is appealed to there. This brings me to the fourth suggestion about why the original story about sympathy is not repeated. His reviewer in the Bibliotheque Raisonée had made fun of the general vivacity-transfer theory, suggesting that it was sheer repetition of Hume's theses about belief that led to the vivacity of his own belief in them.23 So Hume had occasion to rethink, and he expresses some worries about vivacity in the Appendix to the Treatise, published just after that review had appeared. I think he had second thoughts about the entire vivacity theory in the Treatise, perhaps seeing that he had no clear account of what exactly he meant by "vivacity," and insufficient evidence for the claims about its transfer, and what it is transferred from, whether it be taken as degree of assurance, which I think is what he usually meant by the term, or as attention-grabbing power, or as influence on action. So there were at least four good reasons why the Treatise sympathy causal story is not repeated in the later writings: confusion over what counted as impressions, confusion over the definition of cause, confusion over our various self-conceptions, and the role they play in sympathy and also in the blocking of sympathy, and confusions over vivacity. The last two problems concern only the second stage in the sympathy-process, leaving the account of the first stage quite salvageable, though maybe in need of restatement, once the concept of an impression is phased out, and "cause" is re-defined. In addition is the possible overemphasis on reliance on our own experience, given how much he believed the appeal of fine literature, especially tragedy, was due to its enlargement of our experience of the passions, in particular of pity and terror. I think that possible defect could easily have been corrected, and was, in his dissertation on tragedy, though the word "sympathy" is not there used. But what of the other problems?

AW: One could perhaps say that we start with our own experiences but that imitation, reading and conversations can enlarge not only our knowledge but also our passion-repertoire by leading us to new experiences as well as new beliefs. If we additionally grant that one can sympathize in various ways and that the conversion [End Page 76] of an idea into an empathic feeling is only one of them, it looks as if Hume could have maintained his account of sympathy without running into contradictions with his confused account of the self, and the role it plays in sympathy. He could do so by making explicit that sympathy does not require empathic feelings that result from relating ideas of another's emotions to the perception of self.

Furthermore, as Hume allows, sympathy is not impartial. He suggests that the intensity of sympathy-generated feelings depends on the closeness of our relations to those with whom we sympathize. Thus he writes that "we are most uneasy under the contempt of persons, who are both related to us by blood, and contiguous in place" (T; SBN 322). This seems to entail that sympathy biases us and provides us with concerns only for those who are close to us. This problem is removed if ideas are taken to be sufficient for prompting the moral sentiments. Hume writes: "Relations are requisite to sympathy, not absolutely consider'd as relations, but by their influence in converting our ideas of the sentiments of others into the very sentiments" (T; SBN 322). Hence, relations matter only with respect to the conversion of ideas into impressions. So the fact that someone is a stranger to me will not prevent me from forming ideas of her pains and pleasures and from taking an interest in her fortune; it will only prevent me from sharing the feelings that I ascribe to the stranger. The objection that sympathy is biased would thus be neutralized.

This becomes particularly clear if we consider that ideas involved in processes of sympathy respond to relations of resemblance. As I said before, in idea-based processes of sympathy perceptions of third-person behavior trigger ideas of a mental cause. These ideas are caused by associations that proceed upon the model of past experiences: either the experience of a constant conjunction between my own passions and behavioral expressions or, as you have stressed, experiences with other persons' behavioral expressions and their reported passions. In both cases it is the relation of resemblance that leads me to think that we both have the same kind of experiences relating to the same kind of behavioral and facial expressions: either the resemblance between the feelings the other describes and the feeling I sometimes experience myself or the resemblance between her observable expressions and my own performances. The minimal condition upon which one can sympathize by forming ideas of passions that derive from one's own previous experiences here turns out to be resemblance. Since for Hume it is clear that human beings do resemble each other—he tells us that "nature has preserved a great resemblance among human creatures," a resemblance which is "very remarkable" because "it preserves itself amidst all variety" and "must very much contribute to make us enter into the sentiments of others, and embrace them with facility and pleasure" (T; SBN 318)—it follows that we can sympathize with every human being. Of course this conclusion holds only if we accept that genuine processes of sympathy can operate by invoking ideas about other persons and need not result [End Page 77] in empathic feelings, of which the intensity depends on further relations than mere species resemblance.

AB: Hume does in the Treatise say not merely that sympathy is a very powerful principle, one that "produces" our sense of morals (T, SBN 577–78), but later, after considering the objection that sympathy is biased in favor of those close to us, he says that we find it "commodious" to sympathize with those who have any commerce with the person whose character we are judging, if we want to have a fixed standard of moral merit. He goes on to say how much less lively this sympathy will be than if it were our particular friends we were considering (T; SBN 584). His wording here suggests that you are right to take it that it is only the first stage in the full sympathy process that is needed for morality, unless it is our friends (or enemies) we are judging. But it surely must be more than just a belief about the other's state of mind. It must also involve understanding the other's feelings, and an implicit realization that one would feel that way oneself, if in the other's shoes.

And let us hope that something from the second stage of the full sympathy process can also be salvaged, some version of the original Book 2 claims about our near automatic fellow-feeling with those around us. For many of us find wisdom in the Treatise story, and we may prefer it to, for example, Adam Smith's later account, which requires a judgment about propriety to mediate any real sympathy, so we need to decide whether or not another deserves our sympathy, whether her shoes are clean enough for us to step into, before letting her have any sympathy. (Smith could not, like Hume, be said to have prophesized the discovery of mirror neurons, and not only because Hume came first. Even if Smithian sympathy, when it occurs, does involve mirror neurons, their operation does not wait on a judgment of propriety, and propriety neurons are yet to be discovered.) Hume kept thinking about sympathy, and wrote to Smith in July 1759 about Smith's account, in his Theory of Moral Sentiments, where sympathy consists of imagining oneself in others' shoes, as long as those shoes are respectably clean, and the other person sufficiently stoical. In the letter, Hume repeats his own Treatise view that "the sympathetic passion is a reflex image of the principal,"24 so if we sympathize with suffering patients in a hospital, sympathy will not be agreeable. Smith had claimed that all sympathy is agreeable, to give as well as to receive, that we are pleased to find ourselves able to judge another's reaction fit and proper, not lacking in "propriety."25 Were Smith right that sympathy is always something it is agreeable to feel, as well as to receive, Hume writes to him, "An Hospital would be a more entertaining place than a Ball." In this letter Hume still keeps not just his own view that we do sympathize with pain, but also something like his original contagion account of sympathy, so sees the "Damp on Company" cast by a depressed complaining fellow to be due to sympathy. Of course if Smith judged the complainer to be "ill-humored," and [End Page 78] without enough excuse for his moroseness, he would refuse him sympathy. I doubt Smith frequented many balls, and may well have disapproved of the gaiety there on show, so for him a ball may have been as disagreeable a place as a hospital. For Hume, "extensive sympathy" with all affected parties is a preliminary to moral judgment, not a reward for moral acceptability. It is, for him, a fact about our natures that we feel with one another, unless conflict of interests kills sympathy. To adopt a moral point of view, we must overlook our own interest, in part because that would block the sympathy we need to have with all parties, or at least the understanding we need to have of their feelings, before reaching any judgment about what is to be approved. "Sympathy is the chief source of moral distinctions" (T; SBN 618). Hume needs sympathy, for his account of morality, since it is its "noble source" (T; SBN 619), so some revised story about how it comes to happen in us is presupposed by the account of morals he gives in EPM. "Here is the most perfect morality with which we are acquainted: here is displayed the force of many sympathies" (EPM 9.11; SBN 276). So we still are likely to return to the Treatise, even if we have to clean up, complicate, and perhaps weaken the account given there. After all, we would be emotional wrecks if we had to really feel with everyone affected by any moral matter, such as genocide, or rape, that we seriously consider, or even to feel with everyone in a hospital ward. Understanding all parties' viewpoints may be the most that is needed. To feel with the rape victim would be very disturbing, even traumatic, and Smith may be right that we refuse to let ourselves feel any sympathy with the rapist. And as for that "sympathy with the public interest " which Hume says is needed to approve of conformity to useful social artifices, it can scarcely be construed as fellow-feeling with that abstract thing, "the public interest." Nor is it best construed as sympathy with the public servants whose task is to look after it, since they may be corrupt, pursuing their own or their party's interests. Sympathy with the public interest is best construed simply as appreciation of its importance.

As for which feelings a decent person cannot sympathize with, that is a question Hume does not really help us to answer. I do not think he would agree with Smith's claim that we sympathize only with stoically "deadened" feelings, brought down to a pitch the judicious spectator is willing to sympathize with, that we never sympathize with extreme or violent emotions, only those with a seemly mediocrity.26 Hume thinks women more prone to sympathy, and that they "contract a kindness for criminals who go to the scaffold, and readily believe them to be uncommonly handsome and well shap'd" (T; SBN 338). But women are the weak sex,27 and Hume does not second their sentimental softness for criminals, which in any case does not imply any extenuation, on such women's part, of the crime the prisoner is to be punished for.28 Hume thinks we may hate someone for his crimes, but even there he suggests that it is a case of "double sympathy," where our sympathy with the victim, and our anger at the criminal, drowns out any [End Page 79] sympathy we feel for the criminal. It is a case of weak sympathy for the criminal being drowned out by anger, and by greater sympathy for the victim, not a case of no sympathy at all for the wrongdoer. Hume, unlike Smith, does seem to think that nothing human is alien to each of us.

There is another problem about seeing fellow-feeling as "a reflex image of the principal," when sympathy does occur, and that is that the sympathetic feeling does not seem always exactly to correspond with, be an enlivened version of, what we get from the first step in sympathy, nor need it be a reflex image of the original. What the first step yields is a belief about someone else's sentiment, and possibly about that sentiment's causes. Suppose you feel sympathetic sorrow with me when I am grieving over the death of my cat. Your belief will be, "She is grieving over her cat's death." Need your sympathetic grief be grief that my cat has died? Or is it enough if you feel sad that I am sad? Do the causes and objects of our passions get included in the sympathy we hope for, from others? If you never knew my cat, merely know what it is like to lose a pet, then I can scarcely expect you to grieve over my cat's death. But if you have been insulted, and I feel sympathetic anger with you, I must feel anger at the insult to you, certainly not anger at your anger. My belief, which prompts my sympathetic anger, is "You feel anger at his insult." Hume says the second stage of sympathy involves the conversion of an idea into an impression. The impression which results will be "anger at his insult" with the reference to your anger dropped out, although it is indeed a "cause" of my sympathetic passion. Such derivative passions can have not merely two objects (as Hume says pride and humility do) but two causes, what alerts us to the insult, namely your anger, and the insult itself, which is the cause of your anger. If we describe the sympathetic state as "anger at his insult, which angered you," to incorporate my awareness that your anger had a causal role in producing mine, by sympathy, we do not then have a very good match between idea and impression. The idea is about your anger, the impression is a duplication of it, but with a mentioned double cause. This incorporation of your cause of anger in the cause of my derivative sympathetic anger seems plausible for anger, but not so plausible for sadness, or most of the cases where the contagion account of sympathy makes best sense. If cheerfulness or moroseness is catching, we do not thereby catch each other's reasons for good cheer or moroseness. So just how much of the original passion's "causes and objects" gets into the sympathetic feeling seems to vary from case to case, and perhaps that is why Hume says some but not all sympathy requires us to go "deep" into another's mind.

These worries about the causes or "intentional objects" of the original feeling, and their relation to those of the sympathizer, and to the content of her belief about the other may constitute another problem with the official Treatise account. But maybe the difference between sympathetic anger and sympathetic sorrow just points to the many different types of sympathy which Hume recognizes, some [End Page 80] deep, and reflecting, with due allowance for the first-second person switch, not just the emotion type of the original, but also its causes. We could put it this way: the ideas we get when we believe someone else is angry, or sad, need not be exactly the same ones that the other has of the cause of her own anger or sadness. And the sympathizer always has an extra cause of her fellow-feeling, namely her awareness of the other's feeling, which triggers the sympathy.

It may be that some passions, those that are what Hume calls "appetites for evil," which aim to take pleasure in another's pain, are subject only to deep sympathy, when the sympathizer shares the other's sense that they have good reason for anger, for resentment, for hatred, as well as being aware that the other has such an appetite for evil. Bar-room brawls aside, anger is not, like moroseness or cheerfulness, catching. Adam Smith thinks we rarely sympathize with such "unsocial passions." If you express anger at me, I may come to feel it at you, but I may not. I may feel apologetic, not angry, at causing you anger. And I may not share your anger at some third person, but rather sympathize with the target of your anger. Certainly if I walk into a room full of angry people I do not automatically become angry myself. So Smith may be right that some passions are prima facie less apt to be seconded by sympathy than others. Joy, sorrow, fear, hope, can be contagious, but resentment, contempt, and malice spread less readily from person to person. (The good-natured Hume, however, says he knows hatred and resentment more through sympathy than from his own "natural temper and disposition" [T; SBN 317].) This would explain why, when we do sympathize with such negative passions, we must first have a fairly deep understanding of their causes. Still, Hume himself did not restrict the possibility of some sort of sympathy to some subset of the passions, so any cleaned up version of the sympathy mechanism would do well to be vague about how much of the other's state of mind gets duplicated in the sympathizer. Anger requires quite a lot, but most of us can sympathize to some degree with some cases of anger, and maybe of hatred. Only voyeurs may be able to sympathize with the lust of others, let alone with the rapist's desires. The easily contagious passions can spread without a deep sharing of appreciation of their causes. And even for these, such as sadness and joy, it might be best for the anatomist of sympathy not to commit herself to any thesis about the sympathetically transferred feeling "mirroring" the original, in anything other than its emotional type, but to leave the details of how much gets transferred from mind to mind as vague as Hume leaves them, in EPM. And wise also to allow for several sorts of sympathy.

AW: As already pointed out, sympathy is unquestionably more varied and complex than Hume suggests by telling the simple story of ideas that are turned into corresponding impressions if these ideas are related to the perception of the self. But even if we accept this simplifed story as the standard case of sympathy, it is [End Page 81] clear that beliefs about another person's situation as well as beliefs about oneself do matter: they matter not only after a certain passion has been formed but right from the beginning.29 For instance, the belief about myself as someone who belongs to this group rather than that group matters, for it determines how close the relation between myself and the other will appear to be and will thereby decide how much liveliness the idea of the self can pass on to the idea of the other's passion. Hence, my belief about who I am matters with respect to the question of whether I develop moral sentiments upon the belief in the other person's emotions, a belief that I can have without considering the other person in relation to myself, or rather upon empathic feelings, something which would require me to think of myself in relation to the other.

Beliefs about oneself, one's mental and physical properties and position within one's social environment—beliefs that influence the respect in which I take myself to relate to other persons, and so influence whether and how intensely I engage with this or that person's emotions—need to be distinguished from beliefs which are the product of processes of sympathy. Shared beliefs can belong in the latter class: they emerge if sympathy affects me in such a way that my idea of your belief becomes the belief itself. Sympathy can also lead to beliefs that I form about you, but that we do not need to share. This is the case if I form ideas about your feelings and these ideas acquire a liveliness that distinguishes them, on the one hand, from the mere conception of your feelings and, on the other, from the feeling that I would have if my idea became lively enough to qualify as an impression. The reason for this is that for Hume beliefs are nothing but a certain manner in which we conceive our ideas: "Belief or assent . . . is nothing but the vivacity of those perceptions they present" (T; SBN 84). An idea of another's sorrow which is not lively enough to qualify as an impression of sorrow, but which is too lively to be a mere idea would therefore qualify as a belief: that is, as the belief that another person feels sorrow. The belief's liveliness would primarily stem from the causal process which underlies its formation and not merely from the idea of self: "The transition from a present impression, always enlivens and strengthens any idea. When any object is present the idea of its usual attendant immediately strikes us, as something real and solid. 'Tis felt rather than conceived" (T App. 9; SBN 627). If we consider this, it seems to me that the question of how strongly we engage with another's mind can best be answered in terms of the liveliness that pertains to the conceptions, beliefs, and emotions by which we are connected to the other person.

AB: But it is precisely this dimension of "vivacity" or "liveliness" which Hume seems to realize, at the very end of the Appendix, may be not one dimension after all. He has stopped using the notion by EPM, though there is still an occasional reference to "liveliness." Even the distinction between impressions and ideas is not drawn there. Of course he still keeps some distinction between sensation [End Page 82] and thought, mere ideas and believed ideas, and between beliefs and sentiments, but he does not use the Treatise theory of impression-derived vivacity to explain these distinctions. And he never had given a very satisfactory account of just what impressions of sensation—which vary little in vivacity—and impressions of reflection—which vary a lot more in strength—have in common. Both are supposed to be perceptions which have reached a certain important threshold of vivacity. Both are involved in the full sympathy process, sense impressions of how another looks and is behaving, an impression of reflection at the end of the process, one taken to be a reflex image of what the other is taken to be feeling. Doubts about how exactly it must match the other's feeling, and doubts about the "vivacity" that supposedly attaches in its highest range of degrees to impressions, as well the problem about how we are to get impressions of our own perceptions, to establish any causal conjunctions among them, may have stopped Hume repeating the original account. And if there can be considerable range in degrees of sympathy, though there is less room for degrees of relative vivacity in impressions,30 then sympathy cannot always be a matter of having reflex impressions. Hume may have had several good reasons not to repeat the precise Treatise story about how sympathy operates. But that it does somehow operate to give us awareness of and concern for others' feelings is still a vital part of his account of how our moral sentiments come to be possible.

AW: In EPM, Hume discusses sympathy mainly with respect to our ability to approve of moral qualities. Thus, although claiming that "we enter, to be sure, more readily into sentiments, which resemble those which we feel every day," he adds that "no passion when well represented can be entirely indifferent to us; because there is none, of which every man has not, within him, at least the seeds and first principles" (EPM 5.30; SBN 222). Hume's point here seems to be that humans are capable of being affected by the fate of other persons, be this by having an exact representation of the passion in question or by merely forming "lively conceptions" (EPM 5.43; SBN 230) of it. This seems to suggest that a good way of understanding sympathy, and even the Treatise's account of sympathy, is to reject the view that the only way in which Hume takes us to be able to connect to other persons is by having feelings, feelings that exactly mirror the passions of other persons. Morality may well be a matter of our sentient natures, but this does not entail that we need to feel the feelings of all those for whom we care; beliefs about their situations, and about their opinions as well as their emotions, are often suffcient to affect us emotionally and to create an interest in their well-being.

Hume does indeed describe sympathy as a form of communication; however, reflection on the origin of the ideas that sympathy sometimes turns into feelings has shown that by processes of sympathy we re-activate feelings that we have once experienced ourselves. If we ascribe these feelings to other persons, we therefore [End Page 83] cannot be sure that they really feel the way we think they are feeling: we only conjecture. Of course, given a general resemblance among human nature it is very likely that in most of the cases my feelings and subsequent ideas will resemble the experiences others have in similar situations. It can therefore be assumed that I am not entirely mistaken if I associate ideas of mental causes upon the model of my own past experiences. But still it must be clear that sympathy is not an instrument that allows us to enter magically into another's breast.

Again, this is not to say that we can sympathize only with experiences that we have once had ourselves. Some sort of pain experience is needed in order to form the simple idea of pain by which your specific pain can be represented. If sympathy excluded cases in which we sympathize by entertaining resembling ideas of a certain kind rather than by forming the exactly similar ideas of the very passions another is feeling, we could only engage with very few persons.31 Each of us has a different history, different biological dispositions as well as a personal susceptibility to pain. Many of us certainly cannot fathom the tortures persons endure during times of civil wars; and others may not even notice the pain which someone else is finding unendurable. But despite all their variety, it is clear that we are able to think of the pains of other persons and that we are able to develop concerns for their well-being. Hume can explain this, however, only if we think of sympathy not as an instrument that aspires to give us an exact representation of another person's mental states, but as a mechanism that puts individuals in a position to understand and feel for one another, without guaranteeing that our beliefs and feelings precisely mirror the other's real emotional states. If we focus only on sympathy as the source of empathic feelings, feelings that we take to be those the other is really experiencing, this aspect may be easily overlooked. What matters is that we be able to see things from the point of view of what Hume in the conclusion to EPM calls "the party of humankind" (EPM 9.9; SBN 275). This does require that "the force of many sympathies" (EPM 9.11; SBN 276) operate on us, and the force of many kinds of sympathy. Literally feeling with others has its place among them, but it cannot do all the work.

Annette C. Baier  

Annette C. Baier formerly taught in the Department of Philosophy at the University of Pittsburgh, and is now retired, and an associate in the Department of Philosophy at the University of Otago. E-mail: ann.kurt@xtra.co.nz.


This conversation began at a Hume Day at the University of Otago on October 8, 2008, organized by the Early Modern Thought Research Cluster, and we thank those present then for helpful discussion. We are grateful for suggestions of Michael Gill and an anonymous reviewer for Hume Studies.

1. See Kate Abramson, "Sympathy and the Project of Hume's Second Enquiry," Archiv für Geschichte der Philosophie 83.1 (2001): 45–80. [End Page 84]

2. Reference is to section number and paragraph number in the Clarendon edition of Hume's An Enquiry into the Principles of Morals, ed. Tom L. Beauchamp (Oxford: Oxford University Press, 1998) followed by page number of the edition by L. A. Selby-Bigge (Oxford: Oxford University Press, 1894).

3.  References to the Treatise are to David Hume, A Treatise of Human Nature, ed. L. A. Selby-Bigge, 2nd ed., revised by P. H. Nidditch (Oxford: Clarendon Press, 1978), referred to as "SBN" and cited by page number, and to the edition by David Fate Norton and Mary J. Norton (Oxford: University Press, 2002), abbreviated as "T" and cited by Book, chapter, section, paragraph.

4. Árdal defends the view that Humean passions are simple, although he admits that it is incorrect to speak of some of our passions as simple impressions; for a criticism of Hume's understanding of passions as simple impressions see Páll Árdal, "Hume and Davidson on Pride," Hume Studies 15.2 (1989): 387–94 and Donald Davidson, "Hume's Cognitive Theory of Pride," Journal of Philosophy 73.19 (1976): 744–57. Passmore denies that Hume takes passions to be simple. Resemblance, as Passmore observes, entails that passions have parts, some of them being identical and some different to parts of other passions. Hume thus would suggest passions to be complex rather than simple impressions; John Passmore, Hume's Intentions (Cambridge: University Press, 1952), 124.

5. Mercer makes this point and has been criticized for this by Taylor; see Phillip Mercer, Sympathy and Ethics (Oxford: Clarendon Press, 1972), 34 and Jacqueline Taylor, Reflecting Subjects: Passion, Sympathy and Society in Hume's Philosophy (forthcoming).

6. For details concerning the shift in perspective which distinguishes processes of sympathy from ordinary processes of causal reasoning, see Anik Waldow, "Hume's Belief in Other Minds," British Journal for the History of Philosophy 17.1 (2009): 119–32. For discussion of differences that arise from the mechanisms' susceptibility to changes in the perceivable environment, see Waldow, "The Belief-Grounding Function of Sympathy," in Reading Hume's Treatise, ed. Peter Kail and Marina Frasca-Spada (forthcoming).

7. For a disagreeing view, see Mercer, Sympathy and Ethics, 32. Gerald Postema, "Cemented with Diseas'd Qualities: Hume on Sympathy and Comparison," Hume Studies 31.2 (2005): 249–98 tends to support Mercer in taking Hume to mean that comparison can prevent any sympathy, not just cut it off.

8. Reference is to section number and paragraph number in the Clarendon edition of Hume's An Enquiry Concerning Human Understanding, ed. Tom L. Beauchamp (Oxford: Oxford University Press, 2000), followed by page number in the edition of L. A. Selby-Bigge, Hume's Enquiries (Oxford: Oxford University Press, 1894).

9. Annette C. Baier, Death and Character (Harvard: University Press, 2008), part 2, passim.

10. See Alvin Goldman, Simulating Minds: the Philosophy, Psychology, and Neuroscience of Mindreading (Oxford: University Press, 2006), R. M. Gordon, "Sympathy, Simulation and the Impartial Spectator," Ethics 105 (1995): 727–42, and Susan Hurley and Nick Chater, Perspectives on Imitation: From Neuroscience to Social Science, 2 vols. (Cambridge, MA: MIT Press, 2005). [End Page 85]

11. Hume first published this as "Of the Passions" in his Four Dissertations in 1757. Then in 1758 he changed its title to A Dissertation on the Passions and had it appear between his two Enquiries. David Immerwahr edited a Thoemmes Press reprint of the Four Dissertations in 1996, and now there is the Clarendon critical edition of two of these dissertations, Dissertation on the Passions and A Natural History of Religion, ed. Tom L. Beauchamp (Oxford: Oxford University Press, 2007).

12. See T (SBN 387) for the first sense and T (SBN 586) for the second.

13. Postema, "Cemented with Diseas'd Qualities," 274–6 tries to give Hume's explanation of just when this happens.

14. For instance see T (SBN 369).

15. Shaftesbury used the term for the advance sense that pregnant animals have of what their offspring will need, when they prepare a nest.

16. G. E. M. Anscombe, Intention (Basil: Blackwell, 1968), 13ff.

17. Robert Burns, "To a Louse." The last verse of this poem about a louse on some fine lady's fine bonnet, in church, a lady sitting in the pew in front of the poet, runs " O wad some Power the giftie gie us, to see oursels as ithers see us. It wad frae mony a blunder free us, an foolish notion."

18. Adam Smith, The Theory of Moral Sentiments, ed. D. D. Raphael and A. L. Macfie (Indianapolis: Liberty Press, 1976), 11.

19. Smith, Theory of Moral Sentiments, 317. Smith makes this point to show that we really do get outside ourselves, by sympathy, so views that base approbation on sympathy are not "selfish systems."

20. Smith, Theory of Moral Sentiments, 30.

21. "Hume's Post-Impressionism," in Annette C. Baier, Death and Character, 237–56.

22. Postema, "Cemented with Diseas'd Qualities," 279–88, tries to make Hume's view here comprehensible, but I think Hume concedes in EPM 6. 33n34 (SBN 248) that he has left it "unaccountable" why sometimes we feel with others, at other times feel malice and envy.

23. The review is to be found in Early Responses to Hume: Metaphysical and Epistemological Writings, Volume 1: Eighteenth-Century Responses, ed. James Fieser (London: Thoemmes Press, 2000), 44–63.

24. Letter 165, in The Letters of David Hume, ed. J. Y. T. Greig, 2 vols. (Oxford: Oxford University Press, 1932), 1: 169.

25. See Smith, Theory of Moral Sentiments, 13–16 and 43–50. Smith revised some of his claims in the 2nd edition, in response to Hume, and adds a note referring to Hume's criticism on page 46.

26. Smith, Theory of Moral Sentiments, 27. He had earlier said that we so much wish for "concordance" with our sentiments, that we put ourselves in the place of the hoped-for sympathizer, and judge how far he will be willing to enter into our feelings, "abating their violence" (22) when we judge it necessary, to obtain sympathy. [End Page 86]

27. Hume is expressing a gender stereotype here, one which was fairly standard. He may be influenced by Malebranche, who thought women particularly prone to sympathy. He believed that what they witnessed when pregnant was communicated to the child in the womb. He cites Sir Kenelm Digby's claim that James I of England had a fear of naked swords, because his mother, Mary Queen of Scots, was pregnant with him when she witnessed the murder of Rizzio, done with a sword. See Recherche de la Vérité 2, 7, 6 (Paris: Pleiade Press, 1979).

28. In his History of England, Hume relates how Queen Philippa interceded with Edward III to prevent the execution of the six volunteer burghers of Calais. They were not criminals, but heroes, but were dressed as condemned criminals when the queen saw them on their way to the block, and "threw herself on her knees before him [King Edward] and with tears in her eyes begged for the life of these citizens." Far from despising this womanly intercession, Hume says Philippa saved Edward III's memory from infamy. Hume, History of England, 6 vols. (Indianapolis: Liberty Classics, 1983), 2: 258. Hume includes this act of Philippa's in his index, and clearly believes not only that the Calais burghers were heroes, but that Philippa is a heroine.

29. Taylor, Reflecting Subjects, forthcoming.

30. Hume does say that some sense impressions can be "faint and low" (T; SBN 2) yet above the threshold needed to count as impressions, and he clearly thinks passions can vary in strength.

31. Árdal claims that cases in which I fail to feel the precise pain you are feeling do not qualify as cases of genuine sympathy, but as cases of pity. See his discussion of "pity" in Páll Árdal, Passion and Sentiments in Hume's Treatise (Edinburgh: University Press, 1966), 50. [End Page 87]