1 Shoemaker on emotions and the appraisal of quality of will

Shoemaker’s argument with regards to the quality of will approach is an attempt to refine and defend a Strawsonian view of moral responsibility. Strawson puts forward a sophisticated form of compatibilism, and, although Shoemaker is not interested—in this book—in defending compatibilism (p. 9), he is interested in the insights with which Strawson makes his case. Three key elements of Strawson’s position make up what Shoemaker calls the ‘quality of will’ approach: first, the claim that moral responsibility amounts to nothing more than liability to certain emotional reactions, and their associated action tendencies; second, that the relevant emotions are oriented around expectations of goodwill, and are occasioned when there is a lack of the expected goodwill; and third, that the object of moral responsibility is therefore the quality of will, or the conformity of the will to expectations of goodwill (pp. 7–8).

Shoemaker argues that Strawson’s picture gets into trouble when it is applied to ‘marginal cases’ of responsibility. In these cases, agents do have a quality of will (p. 9), and although we do suspend some forms of responsibility-involving emotional reactions, we do not simply adopt the objective perspective. Rather, there are some responsibility-involving emotional reactions that we continue to have towards marginal agents. Shoemaker doesn’t think that we should abandon the quality of will approach in the face of this problem: rather, he thinks that we should refine it by going pluralistic (p. 17). He agrees that moral responsibility amounts to liability to emotional reactions that are oriented around expectations of goodwill. But he argues that any monistic interpretation of ‘quality of will’ fails to provide ‘an adequate interpretation of our ambivalent responsibility responses to several marginal cases’ (p. 14). He therefore claims that there are three different types or ‘families’ of emotion, oriented around different forms of ‘expectation of goodwill,’ and therefore having different conditions of fittingness (pp. 17–18). Emotions ‘fit’ particular objects or situations, in the sense that they are responsive to and appropriate to those situations; and they dispose the agent to certain kinds of action-responses. It is thus through their characteristic objects and through their associated action-tendencies that we individuate and ascribe emotions (p. 21). We appraise people in terms of their characters (or deep selves), their judgements, and their attitudes, and this is because we have three families of emotional response that take these three different qualities of persons as their objects. These forms of appraisal correspond to three versions of responsibility—attributability; answerability; and accountability.

The three families of emotion in which Shoemaker is interested for the purposes of responsibility are set out in contrasting pairs, identified in terms of the objects to which they are characteristically directed, and the action-tendencies they typically involve (p. 26). For attributability, the contrasting emotion pair is admiration and disdain. These emotions take character as their characteristic object. The characteristic action-tendency of admiration is emulation, while that of disdain is the desire to be superior to the other. For answerability, the contrasting emotion pair is pride and agent-regret. These emotions take a person’s judgements as their characteristic object. The characteristic action-tendency of pride is a disposition to reinforce one’s own judgements, while that of agent-regret is a disposition to revise those judgements. For accountability, the contrasting emotion pair is what Shoemaker calls ‘agential-anger’ and gratitude. These emotions take a person’s attitudes as their object. The characteristic action-tendencies of agential anger is a disposition to protest against the slight or seek its perpetrator’s recognition of its wrongness, while that of gratitude is a disposition to return goodwill to the other. These emotion pairings are taken to be universal and non-cognitive ‘sentiments’: such sentiments are pan-cultural and encapsulated aspects of our moral psychology, which have an undeniable influence on our conduct, but which are at best partially under the control of reason. Shoemaker thinks that this view of the emotions, as well as presumably being independently attractive (p. 23), captures what Strawson meant in talking about the naturalness of the reactive attitudes and our practices of responsibility.

I find Shoemaker’s pluralism very appealing. I agree with him that there are various modes of appraisal of agents, and that moral responsibility is not all-or-nothing. I am also inclined to agree that this shows Strawson’s dualism between the engaged and the objective perspectives to have been too simple. I also admire Shoemaker’s discussions of the difficulties of holding marginal agents to expectations in the way characteristic of attempting to maintain distinctively human forms of interpersonal relations with them. However, I think his discussions of various forms of ‘marginal’ or ‘ambivalent’ responsibility in Part Two are sometimes hampered by his need to fit them to the theoretical machinery of the ‘tripartite model,’ and its associated sentimentalist underpinnings, developed in Part One. Despite its attractions, there are problems with that machinery. While the pluralism is correct, the categorisation that Shoemaker presents is too cut-and-dried. While there are ways of justifying such a cut-and-dried categorisation—after all, this is what philosophy is often about in its characteristic ways of stipulating ordered schemes and models to fit messy data—Shoemaker can’t avail himself of them because of his understanding of his project as an investigation of the nature of a set of universal, presumably hard-wired, non-cognitive encapsulated emotions. He can’t simply stipulate about the nature of these emotion-syndromes in the way that armchair philosophers are wont to do in interpreting experience; presumably his grounding for claims about the nature of the emotions must be empirical. However, I am not sure that he provides such an empirical grounding. And it is not clear that he has arguments to show why his categorisation is better than alternative ways of parsing this area.

This introductory note of dissatisfaction sets my agenda for this response. First, I will look at some problems with Shoemaker’s categorisation of the realm of responsibility-reactions into three fundamental types of responsibility. Second, I will look at problems with Shoemaker’s conception of emotion. And, third, I will introduce a distinction between two different ways of taking Strawson’s ‘naturalistic’ approach forward, suggesting that these problems arise because Shoemaker has taken the wrong path. I will suggest that we should interpret Strawson’s appeal to ‘naturalness,’ not in terms of the natural sciences, but rather as it is understood in the tradition of natural law, where what is natural is what is constitutive of some inherently good way of living, or mode of relating. This is not because I am a natural law theorist. I am not. But in drawing on the contours of natural law, I point to a neglected way of understanding Strawson’s project. A successful understanding of modes of responsibility and their associated emotions, I will suggest, must see them as forming part of different ethical models of human relations.

2 An alternative to quality of will: the transgression-guilt model

So, let us turn to some problems with the threefold categorisation for which Shoemaker argues. My concern is that the categorisation is too cut-and-dried—and I am not sure what the status is of Shoemaker’s arguments for carving up the terrain as he does. One way of focusing this concern is to ask how Shoemaker would argue for his set of categories as against someone who proposed an alternative vocabulary of responsibility reactions. We can illustrate these worries by looking at one such alternative. My point is not to argue in favour of this alternative myself, but simply to represent it as a viable possibility that disputes or rejects the Shoemaker/Strawson focus on quality of will, and then to ask what Shoemaker can say to show his own way of approaching matters to be superior. The alternative in which I am interested is what I will call a ‘transgression-guilt’ model of responsibility. The key thing that matters for responsibility, according to this model, is the intentional commission of prohibited actions, and for which, once the intentional nature of the action is established, any further question about the nature of the underlying quality of will is irrelevant. The view I am imagining sees the intentional commission of certain actions as causing guilt (the moral state, not the emotion), which then must be expiated through some form of atonement. The focus on intentional action rules out holding people responsible for what they do accidentally. So this transgression-guilt model can accommodate Strawson’s example in which we react differently depending on whether a person stands on your hand deliberately or by accident. But this model denies that examples like Strawson’s show that the target of our reactive attitudes is only quality of will. The focus on this model is simply intentional action: intentional action is sufficient to commit transgressions that incur guilt.

Is this model part of our moral consciousness? The transgression-guilt model of responsibility might be seen in Oedipus, who attracted guilt and pollution by killing a man and marrying a woman who turned out to be his father and mother. The criminal law might also appear to have this kind of structure. In the criminal law, intention and knowledge are essential to liability, but these requirements are construed in minimal terms (it isn’t always necessary that one intended the action under the description according to which it was wrong, or that one knew what one was doing in the demanding sense that one knew that it was wrong); and, for the purposes of determining culpability in criminal law, it is a commonly repeated dictum that it is intention not motive that matters. Is this model covered by one among Shoemaker’s variety of quality of will approaches, in particular answerability, or quality of judgement? Perhaps, at least in the minimal sense that each intentional action might be said to be done for reasons, and hence the assessment of an action as intentional involves some judgement that it was done for reasons that the agent took to support that action (at least insofar as action ‘aims at the good’). But the view I am imagining is unlike Shoemaker’s category—and hence represents a genuine alternative to it—on the basis that it involves no concern with the particular nature of the reasons for which the agent acted. As long as the action can intelligibly be seen as the product taking themselves to have justifying reasons, it can be seen as intentional. Thus, unlike the answerability model, the content of those reasons is irrelevant to whether the agent attracts guilt for having acted in that way. The only question of interest for the transgression-guilt model is the agent’s judgement that this kind of action was permissible (when in fact it was impermissible). As long as it can be established that the agent made this judgement, this is sufficient for guilt without the presence of any other quality of will.

Of course, the criminal law is a special, and somewhat artificial, case; and my Oedipus example might be said to be archaic. But the question is whether there might not be some form of liability—part of our moral consciousness, and our moral phenomenology—that takes this form, as Bernard Williams argued there is in Shame and Necessity.Footnote 1 Williams discusses the example of Telemachus leaving a storeroom door open, giving Odysseus’s enemies access to armour and weapons. Telemachus is responsible, and must take responsibility, for the consequences of this failure regardless of any failure of concern, or regard, or any erroneous judgements shown in the action. And in ‘Moral Luck,’ Williams famously puts forward the example of the lorry driver who hits a child despite keeping to the speed limit and doing everything needful to keep his brakes in orderFootnote 2; Williams insists in this case that some form of responsibility is involved in the characteristic forms of emotional response and attempts at reparation, even though the lorry driver’s quality of will is impeccable. According to the transgression-guilt model, these reactions are what we would expect if it is possible to have committed a transgression simply by virtue of committing a prohibited intentional action. One way in which an argument that such a conception remains part of our moral consciousness might be defended is to look at Judith Jarvis Thomson’s claims about liability in self-defence situations.Footnote 3 For Thomson, one can incur liability to be harmed quite innocently, without any reprehensible quality of will, by having unintentionally threatened a bearer of rights—for instance, by being pushed over a cliff by a villain in such a way that you will kill an innocent person if she does not kill you first. Here we might see echoes of the conception to which Williams draws our attention: that having entered the zone of transgressive action, one changes one’s normative situation, making oneself either polluted in such a way that one must expiate what one has done, or at least liable, in the sense that the moral boundaries normally protecting one have been significantly weakened or waived as a result of one’s transgression. The transgression-guilt view therefore takes something like Thomson’s model of liability and makes it into a model of moral responsibility: perhaps supplementing its case with the thought that, once we have accepted Shoemaker’s pluralism about forms of moral responsibility, there is no principled reason for arguing that the change in normative situation brought about by one’s innocent (in the sense of no-fault) but transgressive act is not a change in moral responsibility.

As I say, it is not my intention to defend the transgression-guilt view, but only to canvass it as a coherent alternative to those who only recognise quality of will. However, it doesn’t seem entirely implausible to think that people have rights to be treated in certain ways, and that we can incur guilt and obligations to repair if we violate those rights, in ways that are characteristic of moral responsibility, even if we don’t implicate our quality of will in doing so. If I am having an angry argument with my daughter about her staying out too late at night, and, partly as a delaying tactic and partly as a distraction, I glance at my smartphone and see that I have an urgent email revealing a damning problem with my work, I might tell her to get the hell out and not to bother me. Say also that this is completely out of character for me: it was simply the unfortunate combination of circumstances that led to my short temper, and didn’t reveal anything in the way of my lasting character traits or regard. Nevertheless, I have insulted her and treated her in a way that I should never have—or, as we might say, as she has a right never to be treated by her father. Some insulting actions, on the view I am imagining, would be transgressions whenever they are performed intentionally, regardless of whether attitudes such as lack of regard underpin it. That might be because people have rights, in virtue of their status, or their place in a relationship such as this one, to be treated with a kind of respect. Respect, to develop the idea, might therefore be construed as irreducible to proper regard. Proper respect might be in part a matter of performing the right actions, and avoiding proscribed actions, regardless of motivation. No matter how out of character, then, I would have performed a disrespectful act: I would have infringed my daughter’s right to certain kinds of treatment. This could be a case in which quality of judgement is relevant, but like the criminal law, the only way in which my judgement is relevant is the fact that I judged—in the moment of decision—that this action was permitted me. Of course, a proponent of the quality of will view might protest that my action, even if out of character, does manifest an attitude towards my daughter, however fleeting. But the proponent of the transgression-guilt view might ask whether the deep explanation of why I am responsible in this case is really down to the fact that I fleetingly displayed this attitude—or whether it is not that the fact that the attitude is manifested in actions that are inconsistent with the way in which my daughter has a right to be treated.

3 How should we argue for the quality of will approach?

My aim here has been to make plausible the possibility that there is a form of moral responsibility that involves the attribution of actions to an agent and incurs liability, including liability to emotional reactions (reactions to those who insult us and treat us with disrespect), but where this does not involve quality of will being the focus of responsibility. I suggest it in the spirit of Shoemaker’s pluralism, not as the only form of responsibility but as one not covered by his schema. It would therefore be a counterexample to Shoemaker’s categorisation. It is not the only possible counterexample. Consider something more character-based. Shoemaker’s own conceptualisation of responsibility for character is very rich, involving an assessment of one’s deep self. But we can imagine a character-based view that is more austerely aretaic. The model I am imagining simply involves an assessment of how good one is at being the kind of thing a human being is, and of doing the kinds of things a human being characteristically or essentially does. It simply looks at one’s capacities and performances, and at the probability, based on those capacities, that you will act, think, and feel well, as a human being ought to, and grades these capacities and performances against an objective scale of human excellence. This view might sound in some ways Aristotelian; it might also sound like the Hume of the Enquiry, where he claims that we assess people simply in terms of how useful or agreeable their traits are for others and themselves. If we think of that assessment in terms of an objective scale for measuring capacities of a human being in terms of usefulness/agreeableness, and of probabilities that you will do things that are useful/agreeable, we get the kind of thing I have in mind. Again this aretaic view does not focus exclusively on quality of will, at least not if this is construed as the person’s ‘deep self’: it would consider the person’s ‘deep self’—their rational evaluations and their emotional cares—amongst other properties in determining how a person is likely to act, think, and feel, and hence what she is like, and how good she is at being a human being. This ‘austere aretaism’ represents another possible counterexample to a quality of will approach. It is an example of responsibility because it concerns a person’s liability to certain reactions, including emotional reactions (on Hume’s view, love and hatred, pride and humility).

Now Shoemaker might argue that the transgression-guilt view and austere aretaism are implausible alternatives to the quality of will approach. However, if he were to do so, he would require a different kind of argument from any that appears in Responsibility from the Margins. The book’s strategy starts from an assumption that quality of will should be our focus, and then defends that approach against the marginal agents objection; but it does not really motivate the quality of will focus in its own right. Therefore, if we were to ask how we should adjudicate a dispute between Shoemaker as a proponent of the quality of will view, and a proponent of either the transgression-guilt view or austere aretaism, it is not clear that Shoemaker offers us any guidance. Now it might be open to Shoemaker to point out that no book can do everything, that it is legitimate to assume the quality of will approach and seek to resolve certain questions that it faces, and hence that it is somewhat unfair to demand that he should have devoted space to motivating the quality of will view against all of its possible opponents. However, if we look at some of the details of Shoemaker’s quality of will schema, there is a similar tendency to present controversial issues as cut-and-dried. For instance, it is not entirely clear what status we should give to Shoemaker’s arguments for the three ‘syndromes’ that he identifies of accountability, answerability, and attributability. Why does he think that there are only three basic forms of responsibility? Why does he think that agential anger takes as its only focus quality of regard? How would he argue against someone who disputed that? What kinds of arguments would he be drawing upon?

In searching for answers to some of these questions, we might ask how, given the rest of the argument, Shoemaker thinks his focus on quality of will, and his threefold categorisation, are grounded. Three possibilities are that it is grounded: (1) by an empirical investigation into human behaviour, of the sort characteristic of empirical psychology; (2) in a kind of first-personal phenomenology of the agent’s perspective; or (3) by normative argument to the effect that the categorisation focuses on those capacities that are most important about human beings and that therefore play a central role in an attractive form of human relations. The empirical form of grounding is most consistent with Shoemaker’s conception of his own project, and this is because of the role he gives to empirically defined non-cognitive and universal sentiments in constituting moral responsibility. An explanation for what I am calling the ‘cut-and-dried’ appearance of the schema might therefore be that Shoemaker thinks that he can bypass the alternative models of responsibility we have just looked at by appealing to his project of universal sentimentalism. He might argue, on empirical grounds, that there are no such universal human emotions that take as their objects transgressions or arête in the sense we have described. However, the book does not pursue this kind of empirical inquiry, and therefore does not deliver this argument (though it may be intended to?). Its method is more like the phenomenological one. However, if we look for a phenomenological argument grounding the adoption of the tripartite model that would explain why only these categories of moral responsibility are plausible, and which could be deployed against the transgression-guilt or the austere aretaic approach, we won’t find one.

Shoemaker might protest at this point. He might argue that the application of the threefold categorisation in Part Two of his book is an attempt to show that our reactions to particular cases are as the tripartite model predicts. However, the argument here is anecdotal, and does not take the form that would be recognised as rigorous in empirical psychology. But isn’t it an investigation into our empirical psychology that his commitments require?

Alternatively, Shoemaker might claim that he motivates the quality of will approach, and his tripartite model, in his claim that we care about the way we are treated by our fellow agents; and, he might say, what we thereby care about is exhausted by his tripartite categorisation. Strawson’s insight is that we care about receiving the good will of others; Shoemaker might say that this involves caring about others’ judgements and attitudes towards us, and their character traits insofar as these affect us. He might claim that all and only these aspects of a person are what we care about when we engage in moral assessment—not those features that are captured by the transgression-guilt or austere aretaism models. Or he might claim that it is only the quality of will approach that can account for the pattern of excuses and exemptions that we intuitively accept. However, I am unconvinced by both of those responses, at least if the grounding for their claims about what ‘we’ care about is an empirical sentimentalist one. My main line of thought is that the pattern of excuses and exemptions that we accept, and our reasons for focusing on quality of will (if we do) are more theoretically informed than simply hard-wired and universal. Of course, we do not necessarily have an articulate grasp of the theory underpinning our responsibility reactions. But to the extent that we do favour the quality of will approach, it is more likely to be because of our grasp of a particular ethically loaded view of what is important about agents, and of human relations, into which responsibility judgements fit.

Thus, an overlooked possibility for defending the quality of will approach is the normative approach that above I contrasted with the empirical and phenomenological approaches. What is wrong with the transgression-guilt model is to be located in ethical reasons of the sort that Martha Nussbaum discusses in her paper ‘Equity and Mercy’: that its refusal to ‘look within’ the agent reveals a lack of humanityFootnote 4; what is wrong with the austere aretaic approach is that its grading approach is either, as T.M. Scanlon says, a bit pointless, and doesn’t fit in with any important human interests (how humans measure up in this aretaic sense is a possible object of contemplation, but not one to which we should devote too much time)Footnote 5; or because, as on Hume’s own view, it is too aggregative and utilitarian, looking at one’s fitness to bring about desired consequences in a way that would squeeze out more egalitarian forms of responsibility, such as those that see us as equals holding one another to laws that we have collectively made for ourselves. Applied to Strawson, the normative argument for the quality of will approach is something like this: the way of seeing and interpreting human action that underpins the reactive attitudes is part of a much richer form of life than the one—if it were possible—characterised solely by the objective attitude. The fundamental argument for a model of responsibility reactions, on this view, is ethical.

4 Shoemaker on emotions as sentiments

My concern, then, is that Shoemaker’s categorisation of forms of responsibility, though admirably pluralistic, is not as pluralistic as it could be, and that at present it doesn’t have the resources to argue for the way it narrows the field. I have further suggested that the grounding most consistent with Shoemaker’s approach would be an empirical one, given his view that sentiments play such a central role in structuring moral responsibility. By contrast, I have proposed that a more satisfying way to ground the Strawsonian approach is normative. Could Shoemaker respond that his empirical approach is, contrary to my proposal, superior, because it rests on a more adequate conception of the emotions? The normative view, he might rightly say, does not deny that emotions are involved in, and partly constitutive of, our practices of moral responsibility. But the normative view takes these emotions to be malleable and essentially penetrable by morally loaded conceptions of what is important about humanity, and how human beings should relate to one another, in such a way that different conceptions of human relations are embodied in different families or syndromes of emotions. By contrast, Shoemaker might argue that the empirical approach he recommends is more in line with a naturalistic approach to the emotions: it is a virtue of his approach, he might argue, that its view of the emotions is in line, for instance, with Paul Ekman’s idea of an ‘affect program,’ that is, as evolved encapsulated mechanisms rather than culturally elaborated modes of being-in-the-world.Footnote 6 However, I am unconvinced that the affect program model of the emotions is the one most helpful to appeal to in thinking about moral responsibility from a Strawsonian point of view—at least not if it is deployed to the exclusion of the normatively loaded model. Of course, it is true that we have an evolved moral psychology; furthermore, it is implausible to think that our psychology is unaffected by the kinds of homeostatic principles that govern our existence as organisms. But our evolutionary inheritance has also clearly left what seems from a normative point of view to be a wide range of possible variation in emotional vocabularies. Therefore, I do not believe that we can rule out the possibility that what we call ‘emotions’ are sometimes more than affect programs, and that they sometimes become vehicles for moral conceptions that vary across cultures and individuals in a manner corresponding to variance in moral orientation. The question is whether, when we study moral responsibility in the way that Shoemaker does, and attempt to explain the varieties of such responsibility and the proper objects of the emotions they involve, it is most helpful to start from the conception of the emotions as encapsulated or to start with a conception of emotions as cognitively penetrable.

I am not sure that Shoemaker has adopted the right conception of the emotions, at least for the purposes of developing Strawson’s project in the way he seeks to do. Shoemaker doesn’t provide a detailed analysis of what sorts of mental states emotions are. But he does tell us something about what they are not. They are not, for instance, states with propositional content. Despite claiming that they lack propositional content, however, Shoemaker is happy to claim that the emotions he is talking about follow patterns of logical relations. For instance, admiration and disdain seem to be discussed as not just psychological but also logical opposites: that is, they are opposites in the sense that they portray their object in contradictory ways, and that it would be more than merely psychologically uncomfortable to hold both attitudes towards the same feature at the same time. Admiration would involve some judgement that the person is high on some scale, for instance, and disdain, that they are low. It follows that one would be contradicting oneself if one experienced both towards the same person, in respect of the same qualities, without having altered one’s view of those qualities. However, Shoemaker cannot avail himself of that explanation of why these emotions are opposites, since he has rejected the idea that emotions have propositional content, or that they are partly constituted by claims about the nature of their objects. Only if they have content that can enter into logical relations can attitudes be logically contradictory. For Shoemaker, though, emotions are non-propositional affective and motivational states that are not accountable to any logical relations. The best that he could do is to say that empirically it is the case for creatures like us that admiration and disdain tend not to co-occur; and I’m not entirely sure that that is satisfactory.

Furthermore, Shoemaker sometimes talks about first- and third-personal versions of these emotions. However, that these are forms of the same emotion is an assumption that it is natural to make only if we accept that emotions have propositional content, and that, in this case, there is a single, intersubjectively accessible content to which there can be a number of different ‘attitudes,’ depending on one’s relation to the content—for instance, depending on whether the attitude is one’s own or someone else’s. If he has really given up on the idea of such propositional content, Shoemaker would have to find an alternative way of explaining the connection between first- and third-personal versions of the emotions, on which he relies.

Another issue where Shoemaker helps himself to cognitivist terminology without explaining how his non-cognitivist view is entitled to it concerns conditions of ‘fittingness,’ a notion to which he appeals widely. If emotions are non-cognitive states, then they may be triggered by characteristic causes; but the idea of conditions of fit is normally thought to go beyond such triggering. It is normally thought to involve reasons for emotions, where these reasons are reasons of warrant, and not merely motivating or explanatory reasons. Reasons of fit do not amount to all-things-considered justifications,Footnote 7 but they do go beyond triggering. They involve some sense of representational appropriateness—appropriateness to the situation, in virtue of how the emotion represents the situation as being. But reasons for warrant go along with the cognitivist picture of the emotions that Shoemaker rejects. So, the question is: how can emotional states that are non-cognitive, and do not represent situations in terms of propositional content, be responsive to reasons of warrant? It might be thought that such states can only be so responsive in virtue of having content that is responsive by virtue of representing or fitting the situation. But then emotions must have some sort of representational content … and hence propositional content? Well, maybe an appeal to propositional content is not needed here, but again, we need Shoemaker to spell out his non-cognitivist alternative in more detail to show how it entitles him to say the things he does.

Of course, it is true that recalcitrance is an objection with which any broadly cognitivist account of the emotions must deal. But Shoemaker leaps too quickly to the conclusion that it is decisive. One obvious response to the recalcitrance challenge would be to suggest that only a narrow range of emotions are encapsulated in the way that Shoemaker claims, and that higher emotions such as those he is dealing with are much more than affect programs, and hence are more responsive to context than he allows. Another thing that a cognitivist might do is to find ‘partners in guilt,’ by pointing out that even beliefs can lag behind evidence in ways that suggest that emotions are not necessarily different in kind from even paradigmatically cognitive states. Yet another response might be to argue that the parallel that those who talk about encapsulation want to draw between emotions and perception only leads us decisively away from cognitivism if we assume a causal theory of perception; a theory of perception more sympathetic to Kant would take our cognitive capacities to be active in perception (and, perhaps, by extension in emotion) right from the start.

If we take a more cognitivist view of the reactive attitudes, it makes it at least possible that those attitudes may embody a certain theoretically informed, ethical, or even ideological view of human beings, their capacities, and their place in the world. On this normative view, the emotions tied to responsibility are not brute hard-wired reactions lacking propositional content, but rather are ways in which human beings actively make sense of their world, implicating claims about the nature of that world and what is important in it. Our range of emotions gives an insight into what we take to be important, and therefore involves a commitment to certain values, and intersects with our more explicit attempts to account for what matters. In holding people responsible, on this view, one experiences emotions that involve treating them as creatures of a certain sort, in relations of certain sorts, and where those relations are in part upheld by the responsibility interactions in which we are engaged. In experiencing the range of emotions that go with holding people responsible, on this view, we commit ourselves to a range of ethical claims, not just about what we hold people responsible for, but how we hold them responsible. What we say about people and how they matter differs if our main reactions to them are admiration and disdain than if they are gratitude and indignation: for instance, the latter involves seeing someone as in community with one, accountable to shared rationally accessible standards or laws, whereas admiration and disdain imply distance. They fit into different forms of human relations. As I have noted, much more must be said about this ethical route to make it plausible. But I hope to have done enough to establish it as an alternative to Shoemaker’s approach that is worth thinking about.

5 Conclusion

In this comment, I have raised a number of concerns about Shoemaker’s adoption of the quality of will approach. I am not sure that the quality of will approach is given an adequate grounding that defends it against alternative models of moral responsibility; and it is unclear what the argument is for the tripartite model. One possibility is that the tripartite model is meant to be grounded in empirical claims about the structure of encapsulated emotions; but I have argued that those empirical claims are not made out, and that regardless it is doubtful whether this is the most helpful model of the emotions to deploy in this context. In contrast, I have proposed that the Strawsonian approach must be defended in ethical terms, by reference to a vision of the value of living together as equals (in some sense) that is embodied in Strawson’s picture of the engaged attitude, and the emotions it involves.