Ingrid takes herself to have been happily married for many years. But unbeknownst to Ingrid, her husband has been faking his love for her all along and is constantly cheating on her behind her back. Suppose Ingrid never finds out about this. Does the fact that her spouse is cheating on her and is only pretending to love her make Ingrid worse off? Many think this is obviously so.Footnote 1 But others disagree. They claim that since she ‘does not know about the deception, it does not hurt [her]’.Footnote 2 Shelley Kagan, for example, insists that cases like Ingrid's ‘appear to be cases in which well-being is affected, but they must not be, for they cannot be’.Footnote 3 So these cases have remained controversial in the literature on well-being.Footnote 4
To see why, we only need to consider the implications. If Ingrid is in fact worse off, then this is a case where well-being is affected by something not experienced by the subject. This would spell trouble for any theory of welfare that includes an ‘experience requirement’ (the view that in order for any factor to affect a person's well-being, that factor has to be experienced by that person).Footnote 5 Unsurprisingly, people's verdicts on whether Ingrid's well-being is affected here or not differ according to what type of theory they hold. Their intuitions, or their reactions to them, seem strongly influenced by their theoretical commitments.
If there were an independent way to assess such cases, a test for well-being that could resolve these disputes, it would have far-reaching implications for the philosophy of well-being. Can there be such an independent test?
A promising idea has been floating around in the literature at least since Nagel's 1970 article on death. Arguing that losing one's mental capacities makes one worse off, he claims that we pity those who are badly off.Footnote 6 The implicit suggestion seems to be that when we have pity for people, this tells us that they are not doing well. Some such connection between ‘feeling sorry’ and well-being is now accepted by many,Footnote 7 but the idea is never fully developed and it is given neither much weight nor attention in the literature.Footnote 8 An idea with that much potential and initial plausibility deserves serious consideration. It needs to be presented in its strongest possible form. This is what this article aims to do.
Along the way, I will discuss some issues which may have kept the idea from having had more impact. My case study, so to speak, will be Brad Hooker's ‘sympathy test’, which he proposes in the last section of an article on the question whether virtue benefits the virtuous agent.Footnote 9 As I will argue, his proposal faces serious, systematic difficulties, rendering it less powerful than it could be, and ultimately unable to do the work he wants it to do. Addressing these difficulties will bring out the strength of Nagel's idea, and allow us to develop a more powerful test.
I will begin by introducing Hooker's test, and show why it does not serve its purpose.Footnote 10 Then I propose my own test. It avoids the difficulties that any test like Hooker's faces, and develops the idea into its strongest form. I then show how it can give a clear verdict on a case like Ingrid's. In section III I deal with some objections, which will further clarify the advantages of my approach.
I. HOOKER'S SYMPATHY TEST
Hooker is trying to argue that virtue does not make the virtuous better off. To this end, he presents the following test.Footnote 11 Take two lives, which are identical except that one of them lacks one particular property; see whether you feel sympathy for the one lacking the property; if you do feel sympathy, then you must think that not having that property is worse for a person than having it. He applies this to a case where one of the two has, while the other one lacks, the property of being virtuous. Hooker assumes that we do not feel sorry for the one lacking virtue. He thinks this is because we do not think having virtue makes a life better.
Hooker recognizes that, as it stands, the argument is not very convincing, since there might be other explanations for why we do not feel sympathy for the person lacking virtue. He proceeds to consider and dismiss a variety of such alternative explanations, including issues of possible limitations on sympathy, possible grounds of sympathy and issues of desert. Most of the problems he deals with by modifying the scenario; others, by offering reasons why they need not apply to his case. Having thus dismissed these alternative explanations, he concludes that the best explanation for our lack of sympathy for the non-virtuous is that we do not think virtue makes its possessor better off.
Hooker deserves credit for taking Nagel's intuitive idea and incorporating it into a formal argument. Although he uses it exclusively to address the question of whether virtue is a constituent of well-being, he ends with ‘the suspicion that thinking about what we would feel sorry for someone for lacking is our best hope as a method of figuring out what we think are the fundamental categories of prudential value’. I agree that the connection between ‘feeling sorry’ and well-being holds much promise and potential to advance the debate in this field. However, the approach Hooker takes – although, or rather because, it is a straightforward development of Nagel's idea – falls short of realizing this potential.
I won't quarrel over the various individual difficulties he considers and will grant that he can deal with most of them.Footnote 12 Instead I will discuss some general problems. Hooker's approach is vulnerable in two ways: in the way he constructs his test, and in the way he employs it. I will address the latter problem first. Loosely put, he asks his test for something about which it is silent. The element of the test that does the main work is the conditional: if one feels sympathy, one must think that lacking the property tested for makes one worse off. It doesn't say that if one does not feel sympathy one is not worse off. Hooker applies it to a case where one does not feel sympathy, so we are left without a verdict on whether lacking virtue leaves one worse off.Footnote 13 So Hooker is left with nothing more than a possible explanation for why we don't feel sympathy in this case.
He could be read as offering an inference to the best explanation. But it is not clear that this would make the argument stronger. It is not a very strong instance of such an argument, since in some cases it is far from obvious that Hooker is in fact offering a better explanation than the one he is trying to dismiss (I will discuss one such point in more detail below). Note that he cannot even rule out some of the alternative explanations he considers – he merely argues that they need not be the right ones. This of course leaves it open for his opponents to reject his result. All the test can add to his argument, it seems, is some rhetorical force.
The other vulnerability is a more serious problem, one that plagues any straightforward application of Nagel's idea. The problem is a reliance on our feeling sympathy in a given case. But whether or not we have such feelings is contingent in ways which can undermine the test. Many philosophers have argued that feelings and emotions generally are too unreliable to be used in anything like Hooker's sympathy test.Footnote 14 Let me present some specific examples of such worries.
When we survey our emotional life, we will find that we are often subject to what has been called the ‘here and now bias’.Footnote 15 If my friend next to me is hurting, my sympathy for him may prevent or overshadow any sympathy I might otherwise have for a casual acquaintance down the street, who is also hurting. These kinds of familiar effects pose a problem for Hooker's test, since they can falsify its results.
A stronger version of this kind of worry is raised by Jesse Prinz, who argues that we may by our very nature be incapable of feeling genuine sympathy for faraway strangers.Footnote 16 We may speak of ‘those poor earthquake victims in Turkey’, but we won't actually feel sympathy for them. If Prinz is right, this will undermine Hooker's test in a slew of cases.
Another common worry stems from the fact that individuals differ widely in their emotional reactions. Where one feels sympathy, the other might feel contempt.Footnote 17 This will stand in the way of an objective application of Hooker's test, since it is liable to yield different results from one person to another.
A third problem is noted by Hooker himself: our sympathy can be prevented by concurrent but incompatible emotions. We may be unable to feel sympathy for someone towards whom we feel blame or indignation, for example.Footnote 18 This opens up a whole range of ways in which his test might fail. Perhaps we are too distraught ourselves to feel sympathy for others (think of Kant's example of the sorrowing philanthropist).Footnote 19 Hooker's argument that such competing emotions (or, rather, the specific competing emotions he discusses) need not block our sympathy offers little consolation here – the possibility remains, and so does his argument's vulnerability.
These are some objections, which any opponent of the sympathy test can raise against it. There are many similar ones. The general point is that there are widespread worries about the suitability of emotions to serve as epistemic guides in the way sympathy does in Hooker's test. This difficulty affects any version of Nagel's idea that does not go beyond felt emotions. In the next section, I will propose my own test, which avoids these problems.
II. A BETTER TEST
I have said that Nagel's intuitive idea can be developed into a powerful, systematic test. A test like Hooker's, however, faces significant problems, which undermine its viability and weaken its persuasive power. In this section, I will present my own version of such a test, one that avoids those difficulties.
In what follows, I will speak of compassion, instead of sympathy. Although they are often used interchangeably, this is not always the case. Consider the Rolling Stones’ iconic ‘Sympathy for the Devil’: it would have been a very different song, had they called it ‘Compassion for the Devil’ instead. And sympathy is used ambiguously; it can have the sense of feeling sorry for other's suffering, or a connotation of agreement or shared sentiment, as when we sympathize with someone's cause.Footnote 20 Another emotion concept in the neighbourhood is pity. We saw Nagel use it to show diminished well-being. But pity is commonly seen as carrying connotations of condescension,Footnote 21 which can muddy the waters when we assess cases, potentially distorting the results of our test. Not everyone will agree with these distinctions,Footnote 22 and if the reader sees no difference here, she can continue thinking of sympathy, where I say compassion. Nothing much hangs on it for my argument.
Now to the problems I discussed in the previous section. The first issue arose because of the way the test was used. Hooker tried to get the result that a certain factor (i.e. virtue) does not make one better off, by arguing that lacking it does not make a person worse off. In order for something like the sympathy test to yield this result, it would have to contain the necessary condition that only if we feel sympathy for a person lacking a certain property is that person worse off for lacking it. But all we have is a sufficient condition: if we feel sympathy, the person is badly off. To avoid this difficulty, we have to stick to cases where a sufficient condition suffices to yield significant results. This may seem to reduce the range of applicable cases too drastically. But as we will see, once modified, the test can solve cases like Ingrid's, with all the before-mentioned ramifications.
More serious issues arise from the contingency of emotions. There are several objections against using emotions such as sympathy in the way Hooker's test does, based on worries about their reliability. In the previous section, I have discussed three of these. The solution I want to propose involves moving from the emotion itself to an assessment of its fittingness. There is some debate in the metaethical literature on how to understand this notion,Footnote 23 but since our project is not metaethical but epistemological, we need not concern ourselves with it here. There exists a clear and widely accepted sense in which ‘fittingness’ is used in the philosophy of emotion that perfectly serves our purpose.Footnote 24 In making use of the term in that sense, we can avoid the problems noted above. Let's see how philosophers of emotion use it, and how it can help us.
What distinguishes emotions from other affective phenomena, such as moods and sensations, is (among other things) that they are always directed at something – their so-called target. When I am afraid of a tiger, the target of my fear is the tiger. Emotions can be assessed for whether or not they ‘make sense’ with regard to their target. It does not make sense to be proud of the sky, to use an example given by Philippa Foot.Footnote 25 This assessment is based on whether the target is seen as having a certain property, often referred to as the formal object of the emotion,Footnote 26 with each emotion having its particular formal object. Fear, for example, has the formal object of being threatening (or posing a threat). In other words, whether or not fear towards a certain target makes sense depends on whether that target is seen as having the property of being threatening. So fear towards a tiger makes sense when the tiger is seen as posing a threat.Footnote 27
When a target in fact has the property that is the formal object of the emotion directed at it, we speak of the emotion being fitting.Footnote 28 So if I find myself faced with an actual wild tiger in the middle of the Bengali jungle, my fear towards it will be fitting, because in this situation the target of my fear (i.e. the tiger) does pose a real threat to me, i.e. it has the property which is the formal object of fear. If, on the other hand, what I am encountering is merely an especially fine work of taxidermy, my fear towards it would be unfitting, since a stuffed tiger does not pose a threat to me. It might still be rational for me to fear it, if I mistake it for a real tiger, in which case my fear towards it would make sense relative to my beliefs. But the relevant question here is not whether it would be rational or irrational to be afraid, but whether it would be fitting, in the technical sense in which this depends solely on facts about which properties the target has. Since a stuffed tiger does not in fact possess the property of posing a threat to me, fear towards it would be unfitting. To use a more clear-cut example, if I am afraid of a photo of a tiger, my fear is unfitting. This is the notion of fittingness I want to employ.
There are other senses in which an emotion can ‘fit’ a situation: it may be morally called for, or it may be in accordance with norms of etiquette, or prudence. In order to mark these distinctions, Justin D'Arms and Daniel Jacobson proposed using the term appropriate for those other senses, reserving fitting for the technical sense introduced above.Footnote 29 To illustrate, joy at a funeral is usually considered inappropriate, but if the deceased was your personal nemesis, your joy may nevertheless be fitting. If you are attending the funeral of an evil dictator, together with his successor, it will be prudentially appropriate to be overcome by grief, but it would not be fitting. D'Arms and Jacobson warn against inferring fittingness from appropriateness (they call this mistake ‘the moralistic fallacy’).Footnote 30 What we are concerned with here is strictly fittingness.
Now let us turn to compassion. When is compassion fitting? Suppose your colleague claims to feel great compassion for Plato. You will probably be puzzled. After all, Plato was wealthy, smart, successful in the occupation of his choosing and recognized for his achievements. Compassion does not seem fitting towards him. If, however, your colleague specifies that she feels compassion because Plato was robbed of his beloved mentor and friend, who was unjustly put to death, it suddenly seems to make sense. Compassion is fitting towards Plato with regard to this terrible event in his life. Generally, compassion seems to fit when someone is badly off, or has something bad happen to them.
Again, it would not make sense if I reported compassion for my neighbour, but insisted that everything was well in her life and she was doing very well. For my compassion for her to be intelligible, I have to specify some circumstances that make her badly off. Compassion is fitting towards people only if they are in circumstances which make them worse off.
This is confirmed by virtually everyone who has written about compassion, from antiquity until today.Footnote 31 To quote Martha Nussbaum, who has written much on this topic: ‘Literary, philosophical, psychological, and sociological accounts are in remarkable agreement in the descriptions they give.’Footnote 32 This broad agreement suggests that we seem to have a good grasp on compassion and when it is fitting: only if someone is badly off.
With this we can return to the test itself. The core premise in it is the conditional which gives us the sufficient condition for knowing when someone is badly off. Making use of what we have learned about fittingness and compassion, we can reformulate it as follows:
If compassion for some person A is fitting in virtue of A's having (or lacking) some property P, then A's well-being is negatively affected by A's having (or lacking) that property P.
Note that, just like Hooker, I am merely offering a conditional, not a biconditional. This yields a sufficient condition for knowing when welfare is affected, but it leaves open the possibility that some things might affect well-being without compassion being a fitting response.Footnote 33 I do not want to commit myself to an analysis of well-being in terms of compassion's fittingness. To defend such an analysis would take further argumentation, going beyond the scope of this article.
The rest of the structure of the test can largely remain as Hooker suggested. We compare two lives, similar in all respects but that one has a property the other one lacks. Now we see whether compassion is fitting towards one of them, in virtue of having (or lacking) that property. If it is, we know that having (or lacking) that property is negatively affecting that person's well-being.Footnote 34
This approach avoids the problems we found for Hooker. Consider the here-and-now bias. If I don't have compassion for the person down the street who is hurting, because I am focused on my friend next to me who is hurting as well, I can still assess whether or not it makes sense for somebody else to have compassion for the other person. I can be emotionally preoccupied with – or simply biased towards – my friend, without this undermining my sense for the fittingness of another's compassion towards someone down the street who is suffering.Footnote 35
Similarly, for the version of this problem raised by Prinz. Suppose he is right and we are psychologically unable to feel genuine compassion for faraway strangers. Of course, people do say things like: ‘those poor earth quake victims’, but (as we probably know from our own experience), such expressions are rarely accompanied by genuine feelings of compassion. And we shouldn't take such expressions to be synonymous with fittingness judgements. After all, they may be prompted by (conscious or unconscious) considerations of appropriateness in one of the other senses. We might feel that societal norms call for some expression of compassion when watching reports of earthquake victims. Even so, we can clearly see that, say, an aid worker's compassion for earthquake victims in Turkey is fitting towards them, without having to feel any emotion towards them ourselves.
As should be clear by now, my test also does not suffer from the issue of individual differences in emotional experience. Whether one or another person reacts with compassion to a given scenario is irrelevant to the question of whether compassion would be a fitting response.
The possibility of competing emotions does not pose a problem for my test, either. Suppose I feel strong resentment towards an acquaintance who has done me some serious wrong. If this acquaintance now gets badly injured in a car crash, my resentment might prevent me from feeling compassion towards her. Nevertheless, if our mutual friend expresses compassion towards the accident victim (whom I resent), I will not falsely judge this to be unfitting. Rather, I might say that I personally cannot bring myself to have compassion, but I can perfectly recognize that my friend's compassion towards her is fitting. Assessments of the fittingness of an emotion are independent in this way from those emotions themselves. My test works where Hooker's test failed. It significantly strengthens the argument, as the following consideration shows.
Remember that there are philosophers who dispute well-being intuitions in certain cases on the grounds of their theoretical commitments. The compassion test was supposed to provide an independent way to assess such cases, but the idea it is based on has not had much impact in the past. We can now see why. The contingency of our emotional reactions leaves it open to deny the verdict of a test like Hooker's by declaring that, yes, we would feel compassion in this case, but emotions are fickle, and according to theory X the person is not badly off. But it would be incoherent to say that, yes, compassion is fitting here, but the person is not badly off. When an emotion is fitting, this has implications that cannot easily be dismissed. For example, we cannot declare that Helga should be proud of what she did there, but there is nothing good about it. If it is fitting for her to be proud of what she did, this implies that there is something positive about it. That is part of what it means for pride to be fitting. And it is part of what it means for compassion to be fitting that the person the compassion is directed at is badly off. The strongest version of an argument from the connection between well-being and compassion will make use of fittingness. That is what my test does.
It is time to demonstrate its potential by applying it to the case I began this article with. Ingrid deems herself in a loving marriage, while in fact her husband has been faking his love and is cheating on her; she never finds out about this. What is under dispute is whether the deceit and infidelity of Ingrid's husband affects her well-being or not. Let us apply the compassion test. Imagine two women, both under the impression of being in a loving marriage. Their lives are identical apart from the fact that one of them has the right impression, while the other one is mistaken – she is being lied to and cheated on by her husband. It seems that compassion is fitting towards the latter, in virtue of her spouse's cheating on her and deceiving her. If we overheard a conversation in a diner of someone telling his friend about his sister being in Ingrid's situation, and he expressed compassion for his sister, surely we would not find this unintelligible – the stranger's compassion is perfectly fitting here. According to our compassion test, it follows that the woman is badly off in virtue of her being deceived and cheated on by her spouse. Since by stipulation there is no difference in experience between the two women, this is a case of someone having their well-being affected by something that does not change their experience.Footnote 36 Thus, it is a counter-example to any theory of well-being which denies this possibility, such as hedonism, for example.Footnote 37
Those who want to resist this conclusion might want to offer an alternative explanation for why compassion seems fitting here. Surely, they declare, there is no deception so perfect that it would not alter the woman's experience at all. There will be subtle differences in the husband's behaviour. He might not be quite as attentive to her in some situations as he would be if he truly loved her; perhaps he will not give as much thought to the birthday gift he selects for her; and so on.Footnote 38 But note how the improved compassion test can deal with this worry. Since what it tests for is compassion's fittingness, we can ask ourselves what makes compassion fitting – the fact that her husband does not really love her and is cheating on her, or the slightly less thoughtful birthday gift and her husband's occasional slight reduction in attentiveness (if we grant them)? Again it seems clear that it must be the former. How could such subtle differences as we allowed for her make the woman a fitting target of compassion? We can make sure by comparing two women again, both of them with loving husbands, where one of the husbands pays slightly less attention and buys slightly less thoughtful gifts. Does compassion seem fitting towards the wife of the latter? Hardly. So it cannot be those supposed subtle differences that explain why we consider compassion to be fitting towards the wife who is cheated on.
I take it that this example demonstrates the compassion test's potential to help us make progress in the debates over well-being. In the next section, I will address some objections, which will further clarify the strengths of my approach.
III. OBJECTIONS AND REPLIES
I have argued that we can learn about someone being badly off by seeing whether compassion is fitting towards them. But in order to define what makes compassion fitting, I relied on the notion of someone being badly off. Some people worry that there seems to be something circular about this. If so, how can such a test do any work?
In order to answer this worry, let me first reiterate that at no point have I tried to provide an analysis of well-being. I am not claiming that what (diminished) well-being is, is for compassion to be fitting. Instead, I am merely making an epistemological claim. When compassion is fitting towards a person in virtue of a certain factor, this tells us that that factor is negatively affecting that person's well-being. Note that I gave as a necessary condition for compassion to be fitting that its target has to be badly off – this is an only-if claim, not an if-and-only-if claim. And here it is not problematic that being badly off appears in the fittingness conditions for compassion in this way. Consider this: when there is smoke, we know that there is combustion occurring. But in order to define when smoke occurs, we make use of the notion of combustion. Yet, there is no problem of circularity here. We can use smoke as a sure indicator for combustion processes. Often it is much easier for us to detect smoke than to detect combustion. In the same way, we can use the fittingness of compassion to check for diminished well-being. Sometimes this will work better than to check for well-being directly (for example, when our direct well-being assessment is influenced by our theoretical commitments).
Perhaps circularity isn't really the worry. Rather, one may doubt whether the supposed assessment of a case for compassion's fittingness is really anything other than an intuition about whether well-being is affected. It might seem that the former is somehow parasitic on the latter. But this seeming would be mistaken. We can answer a question about compassion without thinking about well-being. As is generally the case, we can look at the same information from different aspects. Assessing an emotion's fittingness need not involve thoughts or intuitions about well-being.
An illustrative case was related to me by a fellow philosopher. When she presented her undergraduates with a Dilbert cartoon in which Dilbert is jubilant at the prospect of living without needing other people, because of his new, powerful computer system, the students expressed that they felt sorry for him – they found him to be a fitting target for compassion. When asked whether they thought Dilbert was badly off, many were unsure at first – after all, he seemed happy. But they reaffirmed that there was something that warranted feeling sorry for him. This shows that they did not form a judgement on Dilbert's welfare, and then made a derivative or parasitic assessment of compassion's being fitting. Rather, it was the other way around.
So not only are our assessments of compassion's fittingness separate from our assessments of well-being in a given case, but we can actually learn something from one about the other. In fact, there are many instances where assessing our or another's emotions for fittingness reveals to us new information – information about what we subconsciously assume, for example. Imagine the following scenario:Footnote 39 a woman joyfully dancing in the street, with a puritan man watching her from the sidelines. The puritan shakes his head, muttering: ‘Poor sinner! I pity her’. We instinctively find this laughable. But what exactly are we laughing at here? It is the mismatch between the puritan's sentiment and the situation as it seems to us. He expresses compassion for the woman, but we don't find this fitting – we find it funny. This reveals that we must disagree with his (implied) judgement that the woman is badly off. Our intuitive assessment of the fittingness of the puritan's emotion tells us something about what we think about how well the woman is doing.
This is no isolated case. Just as we can learn something new by drawing out implications of our beliefs, we can learn something new by considering the implications of our intuitive assessments of emotions (ours or those of others). How we assess the fittingness of emotions tells us something about what we think ‘deep down’.Footnote 40
IV. CONCLUSION
In order to make progress in the welfare debate, we need a way to decide whether certain cases depict changes in well-being or not. I have argued that an intuitive idea by Nagel can be developed into a test to that purpose. I discussed a version of such a test proposed by Brad Hooker, and I argued that it is unsuccessful. I then presented my own test, which relies on the claim that if compassion is fitting towards a person due to her having (or lacking) certain properties, then we know that having (or lacking) those properties affects the person's well-being. I have shown how my test yields results in cases of cheating and deception, which have implications for a central question in the literature on well-being. I believe there is more to be learned from the connection between well-being and compassion, but that will have to wait for another occasion.Footnote 41
