2. The Psychological Problem

Before delving into recent philosophical discussions, it may be instructive to say something first about the abundant social science literature on mixed or ambivalent emotions. Two competing paradigms clash there: one which takes the individual's ability to experience conflicting emotions simultaneously to be non-existent or limited, the other which maintains that mixed emotions can, and frequently do, co-occur.Footnote ⁴ A quick glance at some of the empirical research used to underpin these paradigms reveals, however, a somewhat underdeveloped, if not impoverished, conception of what this ambivalence may involve. As in much of general psychological research on the emotions, the focus is on emotional ‘valence’, positive versus negative – or, more mundanely put, on pleasant versus painful emotions. The question then becomes how positive and negative valence can coexist, and the ensuing puzzle one of putative pleasure–pain ambivalence.

Philosophers tend to be sceptical of valence as a criterion for distinguishing between emotions or emotional experiences.Footnote ⁵ Aristotle argued a long time ago – convincingly I believe – that not only can two differently valenced emotions co-occur unparadoxically, the same emotion may involve both pain and pleasure. For instance, although anger is characteristically painful, it is at the same time accompanied by a kind of pleasure: the pleasure derived from dwelling on the possibility of future retaliation.Footnote ⁶ Similarly, he maintains, the emotion of emulation – pain at the ‘apparent presence among others like him by nature, of things honoured and possible for a person to acquire’ – is mitigated by the pleasure of witnessing and cherishing this honoured thing in the admired person.Footnote ⁷ What matters for emotional specification and individuation is, for Aristotle and most of his philosophical successors, appraisals or evaluations – be they understood in perceptual or cognitive terms. Anger thus evaluates a specific state of affairs negatively; emulation, by contrast, evaluates a specific state of affairs positively, as can be seen from the fact that the emulator would rather not acquire the honoured thing than risk depriving the emulated person of it – this, in fact, is precisely what distinguishes emulation from envy. In any case, following Aristotle but pace much of the psychological literature, I understand the psychological problem of ambivalent emotions to be that of conflicting emotional evaluations rather than of the concurrent experience of emotional pleasure and pain.

Despite this shortcoming in the psychological literature, we have much to learn from it. Indeed, I would argue that, as is typically the case in moral psychology, we would benefit substantially from an interdisciplinary approach.Footnote ⁸ One striking feature which has emerged from empirical studies is the commonality of ordinary people's beliefs that they have been affected by mixed emotions. ‘Folk psychology’ is clearly not a misnomer here. In one study, almost 60% of over 200 respondents were able to describe situations in which they had felt both happy and sad at the same time. Typically, the examples given were of the sort with which I started this paper: examples of being out-performed by close friends or siblings in areas of perceived importance. One respondent wrote, for instance: ‘My best friend and I were both up for a part in a play (the same play). She got the part so I was happy for her but also sad that I didn't get it.’Footnote ⁹

Another interesting finding is that the degree to which people take themselves to have actually suffered from – as distinct from merely experiencing – emotional ambivalence seems to be partly dependent upon the readiness of their own culture to accommodate duality. People coming from cultural backgrounds influenced by Confucianism and Buddhism thus seem to possess greater propensity to accept and embrace duality – including dual emotional messages – than those from Western Christian backgrounds. The latter seem happiest when they are able to construe themselves and their situation in a uniformly positive light, whereas the former are equally – if not more – content with mixed evaluations, where negativity is tempered by positivity or vice versa. Age is also a contributing factor, with older adults being better able than younger ones to endure tensions between conflicting emotions.Footnote ¹⁰

Let us now turn to the potentially troublesome cases that remained at the end of Section 1. To what extent do they constitute a psychological problem of ambivalence, a potential source of embarrassment for dominant emotion theories? I think we can ignore Case 3 with impunity. In that scenario, I am still of two minds about what to feel: The loss of the coveted job has not yet sunk in and emotions have not started to conflict, not even tentatively. There is – to put it in behavioural terms – no gnashing of teeth and cheerful smiling at the same time. Case 3 seems to be, first and foremost, a case of indecision and, as Greenspan observes, indecision should not be confused with ambivalence.Footnote ¹¹

In Case 4, I waver between happiness and sadness about the situation in question. But emotional vacillation is not the same as emotional ambivalence. Frequent and rapid sequential changes in one's emotions may indicate something negative about one's emotional make-up or even character: that one is unduly unstable, even chameleon-like. Such a ‘serial model’, as Pugmire calls it,Footnote ¹² does not, however, point to any deficiency on the part of emotion theories: There is nothing logically or psychologically paradoxical about feeling one thing at this moment and something else at the next.

Case 5 is more interesting: I feel sad, but also ashamed about my own sadness. The question may be asked if these are really emotions about the same thing and hence competing. One is, after all, about the loss of a coveted job, the other about one's own emotional reaction. Even if one accepts that some sort of emotional dissonance is at stake here, I do not think it would be rightly described as ambivalence. Pugmire notes correctly that emotions can be ‘layered’.Footnote ¹³ Much has indeed been written lately about so-called meta-emotions: emotions that have other emotions as intentional objects. Typical examples are angstlust and guilt over having succumbed to malicious joy. We can thus experience second-order joy, annoyance or fright regarding a first-order emotion that we concurrently experience, or, alternatively, regarding some long-term emotional disposition.Footnote ¹⁴ For true emotional ambivalence to occur, however, emotions at the same layer or level of emotional attention must be in conflict. I shall have more to say about meta-emotions in Sections 4–5, after considering the moral problem of ambivalence. But for present purposes, Case 5 should not concern us.

Does loving one's friend but hating the fact that she got the coveted job, as in Case 6, constitutes true ambivalence? I doubt it. Case 6 exemplifies a range of cases where the focus is on two different aspects of the same situation: a general aspect versus a more particular aspect (as in this case) or two distinct particular aspects. Something evaluated as good in one way can be evaluated differently in another way, and there is nothing particularly troublesome or paradoxical about that.Footnote ¹⁵ Consider the cases explored in a recent empirical study aimed at recording how mixed feelings of happiness and sadness can co-occur. The majority of participants surveyed (a) after watching the film Life is Beautiful, (b) moving out of their dormitories at the end of school, and (c) graduating from college, claimed to feel both happy and sad.Footnote ¹⁶ Interesting as those findings may be for some purposes (such as exploring concurrent pain–pleasure ambivalence), they hardly tell us much about ambivalent emotions qua evaluations. There is no evaluative conflict – not even an apparent conflict – between being saddened at the horrors of the Holocaust and being moved to joy by the uplifting evil-has-no-power-to-dissipate-hope message of Life is Beautiful, or between being nostalgic about the friends that you may not see again next year and looking forward to returning to your family for the summer. Evaluations involving love, as in Case 6, complicate matters even further as we have good reason to doubt that love possesses the required specificity to constitute an emotion in the first place. It seems more appropriate to think of love as an umbrella term covering a range of different – if interrelated – emotions, dispositions, desires and behaviours. Recall the question posed by the poet Swift:

Persons in love are aggrieved when the objects of their love are hurt, jealous when the beloved ones start to show undue attention to others, joyful when the love is requited and so forth. Any one of a range of negative emotional evaluations of a person may come into conflict with one's love of that person, but if love is understood as a complex and multi-layered attitude rather than as a distinct emotion,Footnote ¹⁷ then the conflict in question will not be that of two emotions directed at the same state of affairs colliding and, hence, not an instantiation of the kind of emotional ambivalence under discussion in this paper.Footnote ¹⁸

Will the same manoeuvre work to obviate the apparent problem in Case 7, where I am at the same time happy that my friend got the job and disappointed that she got it? Hardly, because there two specific emotions collide (not a general attitude versus a specific emotion) and they seem to have the same formal object, namely my friend's job success, evaluated differently (rather than the same state of affairs divided up into different formal objects, and then evaluated differently). There are philosophers who might want to dig in their heels and insist that the two emotions in Case 7, the disappointment and the happiness, do not in fact have the same formal object, as the first focuses on the frustration of my desire and the second on the satisfaction of that of my friend. I have sympathy, however, with Greenspan's slighting dismissal of such nitpicking manoeuvres.Footnote ¹⁹ The two emotions in Case 7 are not just about different aspects of the outcome of the job race, they are essentially about it. So there we finally seem to have a case of pure emotional ambivalence – the psychological problem in a nutshell – or as Pugmire calls it: ‘simple convergence of dissonant emotions on exactly the same thing at the same time’.Footnote ²⁰

Case 7 then cannot be dismissed as the other cases have been. But the question remains if such a case of true emotional ambivalence – involving a normally rational agent – constitutes a psychological problem to which standard theories of emotion fall prey. My claim is that it does not. The reason lies in the essential partiality of emotional evaluations. As Ben-Ze'ev has argued, emotional evaluations are typically moral evaluations of a specific kind: namely, less sweeping in compass, and more personal and particular than non-emotional moral evaluations.Footnote ²¹ This characterisation does not rule out the possibility of more encompassing emotions – say, sympathy with the victims of suffering in the world as a whole – but such emotions are the exception rather than the rule and, in any case, they are not at issue in Case 7. There is no logical or psychological demand (as distinct from a moral demand) on rational agents that they pass an all-things-considered judgement about the content of their emotional reactions towards a given situation. Indeed, such a demand would be absurd. Emotional evaluations have much in common here with aesthetic ones. There is no logical or psychological contrariety involved in finding the Evening Star beautiful and the Morning Star ugly, even after realising that they are actually both the same celestial body, Venus. ‘So do you find Venus beautiful in general or not?’ seems to be quite an inappropriate question to ask in such a context. The same would go for the question ‘But are you in general happy or disappointed about your friend's getting that job?’ if the focal point of the question is to identity some ‘core’ psychological content of one's reaction rather than to demand a morally normative answer – which is quite another thing. Therefore, I agree with Greenspan that the ‘logic’ (and, I would hasten to add, the psychology) of emotions does permit ambivalence:Footnote ²² Because no logical contrariety is involved, cognitive theories of emotion escape unscathed. And for those who endorse perceptual or biological theories, there may well be specific perceptions that are, by their very nature, perceptions of ambivalence, and special facial expressions of, say, bittersweetness. In any event, I do not think that any of those theories has much to fear from the existence of emotional ambivalence – which is not to say that there are no other logical or psychological hurdles over which they may stumble.

Pugmire may seem to disagree with this conclusion since he emphasises the extent to which ambivalent emotions can never be ‘full-bodied’ but, rather, remain ‘inchoate’ and ‘unconsolidated’.Footnote ²³ He casts doubt both on the functional properties of emotional ambivalence and on its relation to action (as emotions are motivators but here motivate, ex hypothesi, to discordant actions). It is not entirely clear how far Pugmire's suspicion of emotional ambivalence goes (especially in so far as his suspicion is conceptual as well as psychological), but as I understand him, he is not rejecting the view that such ambivalence can exist; rather, he is simply making two different empirical claims, neither of which I see any special reason to question. The first is that an ambivalent emotion is likely to be less intense and profound than a non-ambivalent emotion (as the former will be constantly checked and qualified by the other conflicting emotion), and the second that the life of an emotionally ambivalent agent will be more susceptible to psychological dysfunction than that of a non-ambivalent one.Footnote ²⁴ To be sure, ambivalence places extra burdens on us and such burdens can be the straw that breaks the back for some individuals – but this is not to say that extra burdens cannot also be a source of strength and inspiration to others even if they are, perhaps, in the minority. I revisit that issue in Section 5.

D'Arms and Jacobson have identified what they call a major fallacy in recent theorising about emotions: the ‘moralistic fallacy’ of inferring from the claim that it would be morally wrong to feel an emotion that the emotion is representationally unfitting. They consider the word ‘appropriate’ to be a basic source of ambiguity here, and accuse writers such as Greenspan of constantly conflating moral appropriateness or inappropriateness with representational appropriateness or inappropriateness.Footnote ²⁵ I agree with them that Greenspan does not prise the two notions sufficiently apart. In the literature on emotional ambivalence, however, what can be more commonly seen is a reverse ‘moralistic fallacy’: the fallacy of believing that by determining whether or not emotional ambivalence is rationally fitting, one has thereby answered the question of whether or not it is morally appropriate.Footnote ²⁶ Too much energy has, consequently, been spent on the first question when it is actually the second that most of the participants in the philosophical discourse seem to have at heart. Nothing I have said so far in this paper about the rational fittingness of ambivalent emotions goes any distance towards showing that they are, therefore, morally appropriate. Indeed, Pugmire's empirical claims about the practical implications of emotional ambivalence give us good reason to fear that although the ‘psychological problem’ of ambivalence may have been largely exaggerated, a serious ‘moral problem’ still remains.

3. The Moral Problem

There is, strictly speaking, nothing irrational about sitting on the fence when faced with moral dilemmas. But moral theories tend to require more of us than that. For action-centred theories, such as Kantianism or contemporary forms of utilitarianism,Footnote ²⁷ what matters in the end for moral value is doing the right thing. In contrast, virtue ethics focuses on the intrinsic role of emotions in all the moral virtues. What marks virtue ethics as unique is, most distinctly, the claim that, in some cases, feeling the right thing is not only necessary but also sufficient for virtue. Compassion, for instance, has moral worth as such and constitutes a virtue even if one is, for some reason, barred from putting this proper emotion into action. To lead a good human life, one must cultivate the virtues and hence one's properly felt emotions. What happens, however – or rather, what should happen – when, as in Case 7 above, potentially virtuous emotions collide? I say ‘potentially virtuous’ because we can imagine that I deserved the job better than my friend, and pain at an unjust decision is clearly a potential virtue. Surely virtue ethics, of all moral theories, could be expected to provide some guidance as to what would be the properly felt emotion in such cases and how it should be arrived at. In the remainder of this paper, I shall concentrate on the capacity of virtue ethics to solve this ‘moral problem’ of emotional ambivalence.Footnote ²⁸

A common objection urged against contemporary versions of virtue ethics is that they fail to provide across the board guidance in scenarios involving virtue conflicts. To be sure, rejoicing with friends over their successes exemplifies a virtue and so does being duly disappointed when one's own reasonable life plans are (unjustly) thwarted, but how should one feel, all things considered, when one's life plans are (unjustly) thwarted because of a friend's success? There is a strong particularist strain in contemporary virtue ethics that rejects any appeals to general, context-independent moral truths and relies rather on the virtuous agent's intuitive artistry in coping with situational demands. Virtue ethicists of the particularist type thus refuse to adjudicate between the demands of different virtues (which they consider to be of equal or incommensurable value) in moral dilemma situations, in order to tell us either what to do or what to feel. Hursthouse even makes a special virtue out of virtue ethics' refusal to provide us with a decision procedure in hard moral cases, claiming that not telling us what to do there is ‘entirely to its credit’.Footnote ²⁹

Opponents of contemporary virtue ethics consider non-action-guidance to be the theory's main vice. I would add that as disconcerting as Hursthouse's nonchalance regarding virtue ethics' non-action-guidance is, it is even less satisfying when applied to the theory's non-emotion-guidance. There are three reasons for this: First, in the case of conflicting demands of different virtues to act in one way or another, we are typically guided by various ‘signposts’ and ‘traffic lights’ – social norms, contextual conventions and practical considerations – and we often have enough time on our hands to consider the proper course of action. In the case of conflicting demands about what to feel, social and contextual conventions offer much less help – and we cannot simply hold our emotions in abeyance until we decide what would be the proper way to feel.Footnote ³⁰ Second, empirical research shows our actions to be more situation-dependent and more easily swayed by external pressures than our emotions. A person's essential character or ‘moral selfhood’ seems thus more truly manifested through emotions than actions.Footnote ³¹ Third, one reason why so many people are drawn towards virtue ethics in the first place is, quite likely, because of its professed ability to account for the moral salience of the emotions, not only theoretically but also practically. And why should that not also apply to cases where virtue-driven emotions come into conflict? Simply conceding its uncodifiability is of little consolation for those who see in contemporary virtue ethics an avenue for the successful and morally discerning regulation of their emotional lives.

Aristotelian virtue ethics differs from the typical contemporary versions in being more generalist. The right thing to feel and do is not only to be gauged by the details of the given situation and the virtuous agent's particularist insights, but also by general truths about human eudaimonia that must be applied to particular cases. Although Aristotle is at pains to stress that such an application is not easily codifiable and requires considerable attention to situational details, he does not shirk from giving advice about how to solve difficult moral dilemmas – his works are packed with such advice – and he rejects the thesis that the moral virtues are either incommensurable or of equal standing, claiming instead that those virtues are most important which benefit other people most. There is no space here to give a full account of Aristotle's moral generalism, let alone his virtue theory and his nuanced discussion of the psychological and moral aspects of human emotions.Footnote ³² Let it suffice to say that it is not immediately obvious how Aristotelian virtue ethics succeeds in parrying the bullet that its contemporary counterpart simply bites. Recall the conclusion that stood out at the end of Section 2:

Aristotle's treatment of the emotions and virtues yields a number of assumptions that do not sit comfortably together when combined with (a). Let me single out for consideration a set of such assumptions on which a number of recent discussions of the moral problem of emotional ambivalence can subsequently be brought to bear:

Assumptions (a)–(f) are not strictly incompatible, but they give rise to a mystery. We could call it ‘The Mystery of the Missing Motivation’: Let us suppose, for the sake of argument, that the virtuous person's balancing act of general moral truths and situational appreciation yields the conclusion in Case 7 above that, while both happiness and disappointment are rationally warranted there (see (a)), the morally optimal way to feel (see (d)) is to rejoice with the friend. Ex hypothesi (again according to (d)), the virtuous person is motivated to feel in that way. Yet, this person also presumably has in place a virtuous disposition to feel disappointed when a reasonable life plan is (unjustly) thwarted. According to (b), the person cannot help feeling that emotion, and that emotion is at least weakly motivating (see (c)). Nevertheless, being motivationally unified (e), the virtuous person does not need to suppress the non-optimal weak motivation (f). The remaining mystery is, then, what happens to that non-optimal motivation. What may appear at first glance as a rather trivial and technical quandary has significant ramifications for the tenability not only of an Aristotelian solution to the moral problem of emotional ambivalence but of an Aristotelian approach to emotional virtue in general. Although the mystery in question has not been couched previously in the terms chosen above, it is instructive to explore how a number of recent contributions to Aristotelian virtue ethics would be inclined to unravel it.

John McDowell was the forerunner in a series of particularist readings of Aristotle that were launched in the late 1970s and have since aroused considerable interest. According to McDowell's interpretation, someone who has been properly raised in the Aristotelian way has been habituated into seeing the appropriate actions or emotions as worthwhile in the specific way that is expressed by bringing them under the rubric of the concepts of eudaimonia and the noble or virtuous. Rather than furnishing us with a universal blueprint from which virtuous (re)actions can be systematically worked out, in this contrasting picture there is nothing for the virtuous person's grasp of the content of the universal to be ‘except a capacity to read the details of the situation in the light of a way of valuing […] into which proper upbringing has habituated one’. The content of the universal is not isolable, even in principle, from this learnt capacity. There is no subsequent grounding of the universal from the outside either: The content of this conception is ‘fixed once and for all’ in the minds of people who have been well brought up, without the need for further foundation.Footnote ³⁸

McDowell reads Aristotle almost as if the latter were a late-Wittgensteinian: Habituated persons have learnt a language game involving the concepts of eudaimonia and virtue or nobility. What they later add themselves is increased mastery of the game as they engage in the practices of applying the terms to particular instances; the concepts become more discriminating of subtle shades and nuances through sustained use, without the need for any generalist definitions or independent groundings. In a sense, McDowell rejects assumption (b) by excluding the possibility that any potential uncontrollable, loose-canon motivations can remain for the virtuous person. What motivates that person are beliefs only, not desires or emotions (in so far as emotions are conative as well as cognitive) – or perhaps, rather, the distinction between beliefs and desires is out of place, to be replaced by ‘besires’. Moreover, in the psyche of virtuous persons, other beliefs and considerations than the virtuous ones into which they have been habituated have been ‘silenced’ once and for all. There is no weighing of contrasting or ambivalent considerations, no balancing acts, no trade-offs. McDowell's reconstruction of Aristotle here is monolithic in two senses. First, because virtue is unified and emotions are constitutive of virtue, they also are unified.Footnote ³⁹ Second, motivation itself is unified and based only on cognitions.Footnote ⁴⁰

The problem with McDowell's solution to the moral problem is that if it is to be considered ‘neo-Aristotelian’, then the emphasis must be squarely on the ‘neo’. In our postmodern age, where Leviathans are out and Lilliputians in, particularist readings of Aristotle strike a chord with many readers. As indicated earlier, however, they have precious little to do with the historic Aristotle, that unrepentant systemiser and lover of general moral truths.Footnote ⁴¹ More specifically, McDowell's two monolithic theses give false colour to anything that might be construed as an Aristotelian virtue ethics, however broadly. First, the virtuous person's relationship to virtue is not as mechanical and unproblematic as McDowell depicts it. Rather, that person commonly faces tragic dilemmas and is even, sometimes, doomed to fail at hitting the virtuous golden mean of reaction or action (I say more about such eventualities in Section 4). Second, it seems to be entirely clear that Aristotle was – to put it anachronistically – a ‘Humean’ with regard to motivation, holding that only desires, not mere beliefs, move us to action: ‘Thought by itself […] moves nothing’, he says; rather ‘the origin of an action […] is decision, and the origin of decision is desire together with reason that aims at some goal.’Footnote ⁴²

Importantly, being a ‘Humean’ about the role of desires in motivation does not necessarily make one a motivational internalist – holding moral judgement to be intrinsically motivating – for the action-moving desire may be thought to be external to the specific judgement.Footnote ⁴³ Aristotle is, in general, however, considered to be both a ‘Humean’ and a motivational internalistFootnote ⁴⁴ – and obviously McDowell takes Aristotle to be an internalist, albeit a non-Humean one. Before turning to Susan Stark's externalist reading of Aristotle, it is worth noting that the application of labels such as ‘motivational internalist’ to thinkers who did not define themselves with respect to such categories is always going to be problematic. That is true even in the case of Hume – typically considered a classic internalist – not to mention Aristotle. If one sticks to the text of the Nicomachean Ethics, Aristotle appears to be speaking as a prototypical internalist. That work has a specific target audience, however, namely people already ‘brought up in fine habits’.Footnote ⁴⁵ It is unsurprising that among those lurk no amoralists – passing sincere moral judgements about the right things to feel and to do without being moved by them at all – but this does not mean that Aristotle took the existence of such persons to be a conceptual impossibility as internalists normally do.

Stark critiques McDowell's neo-Aristotelianism from the perspective of the moral problem of ambivalent emotions under discussion here, and she finds it seriously wanting. Instead, she suggests that we drop assumption (c), that all occurrent emotions are at least weakly motivating. When two conflicting but virtue-based reactions to a given situation are potentially appropriate, as in Case 7, we do not simply want the considerations inherent in the overall less virtuous option to be ‘silenced’ or ‘unrecognised’ and pushed out of view. Such a manoeuvre makes the psychology of the virtuous person seem singularly one-dimensional, not only lacking in complexity and profundity but also immune from the sense of the tragic that permeates any mature human self.Footnote ⁴⁶ Nevertheless, we want that person to experience the morally proper emotion in the end (see assumption (d)) and to be motivationally unified, and hence distinguishable from the continent person (see (e) and (f)). How are these claims to be made compatible? Stark's ingenious suggestion is to reject – on Aristotle's behalf – motivational internalism and maintain instead that, in the case of virtuous agents, they can experience emotions (qua moral evaluations) as yielding normative reasons but without having intrinsic motivational force. The ‘evaluativeness’ of an experienced emotion will be fully expressed in the virtuous person's emotional state. What the quest for overall virtue ‘silences’ in such a person is, therefore, not the non-optimal normative reason – its very non-optimality continues to be experienced as a reason for a sense of loss – but rather that reason qua motivational reason. The difference between the virtuous and the continent is, then, that whereas the former can experience two ambivalent, conflicting emotions but only be motivated by one, the continent continue to be motivated by both emotions, which renders them motivationally disunited. Thus, the continent are forced to continue weighing competing motivational reasons for and against the virtuous course and, in the end, must suppress one.Footnote ⁴⁷

David Carr renounces Stark's suggestion, at least in so far as it is meant to be generated by insights found in Aristotle's own writings. He considers it puzzling why there should be any less difficulty for Aristotelian virtue ethics to accommodate non-motivating non-optimal emotions than motivationally active ones. More specifically, Carr claims that Stark confuses virtuous agents with continent ones, and continent agents in turn with incontinent ones (according to Aristotle's specifications of those respective states of character). If Stark's ‘virtuous’ persons need to ‘silence’ a reason for feeling or acting, albeit qua motivating reason rather than normative reason, then they are really continent according to Aristotle's specifications; and if two or more competing motivating reasons remain active for her ‘continent’ persons, then they are really incontinent. In contrast, Aristotle's continent persons have an unswerving and undivided devotion to the correct way to feel and act: they are described at best as impeccably self-controlled, at worst as efficiently repressed. Notably, Carr himself does not propose a definitive and detailed solution to the moral problem as McDowell and Stark do. What he says indicates, however, that he would want to reject assumption (e) of the motivational unity of virtuous persons, and replace it with a richer but more complex understanding of their virtuous unity – mediated through phronesis: unity which would make room for emotional ambivalence and personal conflict as part of a constructive learning curve toward ever greater virtue.Footnote ⁴⁸

4. Virtue, Continence and Phronesis

In order to subject not only Carr's criticism of Stark's proposal but also his own under-explored antidote to further scrutiny, it is salutary to rehearse briefly first some of Aristotle's remarks about incontinence, continence and (full) virtue.

In Aristotle's stage theory of moral development, ‘the incontinent’ are to be found above the levels of ‘the many’, ‘the soft’ and ‘the resistant’. They have managed to overcome the thrust of the painful appetites that prevent many people at the lower levels from aiming at the good. The incontinent also have correct opinions about what is morally right and what they should do. They are easily overcome by counter-moral pleasant appetites, however, and, owing to a complicated intra-psychic process, fail in such circumstances to do what they themselves think they should do. In other words, the incontinent fail in many cases to abide by proper reason ‘because of too much [enjoyment]’.Footnote ⁴⁹

Above the level of the incontinent are ‘the continent’. The continent have managed to subdue permanently base desires, both painful and pleasant, and are able to do the right thing. They are fully self-controlled, and the non-rational part of their soul diligently listens to reason. Yet such self-control is not an ideal state, because continent persons still possess base desires: they want to act badly, but only force themselves to act as they should. We can see therefore what a tall order full virtue is in the Aristotelian model. At that level are the truly virtuous who ‘find nothing pleasant that conflicts with reason’. Full virtue is only achieved when the desires ‘share in reason’ in the strong sense of ‘agreeing with reason’. The desires and emotions of the virtuous are so constituted (reason-infused) as to allow them to feel and desire in the optimal (medial) way, according to Aristotle's theory of the mean of action and reaction. Notably, McDowell's phrase of reason ‘silencing’ desires does not appear here; only that of desires ‘sharing in’ reason. We can see by this comparison why continence is not considered a virtue at all, but only ‘a sort of mixed state’, the lesser of two evils: a ‘second-best tack’. For while the continent and the virtuous ‘are both the sort to do nothing in conflict with reason’, the continent person has base desires, but the virtuous person lacks them.Footnote ⁵⁰

Carr criticises Stark for confusing the virtuous with the continent and the continent with the incontinent. We can now see that something is amiss in Carr's criticism. As Stark is thinking of cases of ambivalent emotions, we can focus here once again on Case 7 from Section 1 (supposing further, as in Section 3, that the optimal emotional reaction is to feel happy for the friend). Stark would consider a virtuous person in this case to have two active but conflicting normative reasons: one to feel disappointed, another to feel happy. Carr claims that if a person does have those conflicting reasons, then that person is merely continent. But that cannot be so, because a continent person has base desires which do not ‘share in reason’; the desires of the virtuous are, however, ex hypothesi, not base. In Case 7 they are clearly not base because (given that the person is virtuous) both the emotion of disappointment and the emotion of happiness are derived from reason-infused, morally proper emotional dispositions (hexeis). Stark would consider a continent person trapped in Case 7 to possess two conflicting motivational reasons, but finally controlling one by force of will. Carr views such a conflict-ridden person as incontinent: not properly ‘repressed’. But apart from the fact that the person manages to police the non-optimal emotion in the end (which the incontinent would not), recall that what the incontinent suffer from is their lack of resistance to pleasures. In Case 7, however, the person is fighting off the non-optimal emotion of feeling disappointed. And disappointment is not a pleasant emotion.

Although Carr's criticism misfires, a more serious objection can, I think, be lodged against Stark's externalism. Motivational externalists believe that persons can sincerely pass moral judgements without being moved by them. As evaluations are not the same as prescriptions, someone might thus say (and mean it) ‘It is morally right to help this poor beggar’ without having any motivation, weak or strong, to do so – unless the person also possessed an external desire to do what is morally right. The trouble with Stark's externalism is that it is not an externalism about moral judgements of this kind (as mere evaluations) but about moral emotions. But emotions already contain within them – according to most emotion theories at least – a conative as well as a cognitive component. Emotions are thus intrinsically motivating in the way that some other types of moral evaluations may well not be if moral externalists are right.Footnote ⁵¹ And if that is so, one simply cannot be a moral externalist about emotions. The only way to ward off this concern would be to reject the view that emotions have a conative element and understand them as mere cognitions. Some emotion theorists do hold such a pure cognitive view.Footnote ⁵² But that is not Aristotle's view of the emotions – for him emotions are clearly affective and conative as well as cognitive; they are those things on account of which ‘people come to differ in regard to their judgments, and which are accompanied by pain and pleasure’Footnote ⁵³ – so this response would not solve the moral problem of emotional ambivalence in an Aristotelian virtue ethics, as Stark set out to do.

As the difficulty in solving the moral problem persists, it may be helpful at this point to follow Carr's hunch about the possible salvaging role of phronesis. Phronesis is, after all, referred to by Aristotle in connection with his much-discussed thesis of the unity of the virtues. Some interpreters have understood that thesis to exclude the possibility that a virtuous person can face moral dilemmas. But, as already noted, Aristotle discusses such dilemmas in detail. A more liberal understanding of the thesis would be based on the role of phronesis in overseeing the virtues and adjudicating the relative weight of each of them in conflict scenarios. After all, what Aristotle says about the unity of the virtues is simply that when one has phronesis, which is a single state, one has ‘all the [moral] virtues as well’.Footnote ⁵⁴ Susan Wolf works, in a recent paper, on the hypothesis that the unity-of-virtue thesis is indeed about the workings of phronesis. The kind of knowledge that is required for the unity of the virtues, on Wolf's Aristotelian or quasi-Aristotelian conception, is knowledge of what is important. Phronesis yields such knowledge. Because the importance of one item requires knowledge of the importance of everything else against which it may in principle have to be balanced, knowledge of value is essentially unified. Thus persons with such unified phronesis inevitably have their priorities straight. And persons who have their priorities straight will ‘see in a flash what needs to be done, whether it involves rushing into a burning building to try to save a child or stopping along the highway to help an accident victim’.Footnote ⁵⁵

My complaint with this picture of phronesis is much the same complaint urged earlier against McDowell's neo-Aristotelianism: It makes Aristotelian moral virtue seem too monolithic and seamless. A satisfactory account of the unifying operations of phronesis must preserve Aristotle's acute sensitivity to moral luck and the tragic vicissitudes of human life – be it lived by the virtuous or the non-virtuous.Footnote ⁵⁶ With regard to ambivalent emotions in particular, it is simply unrealistic to say that it will come to the virtuous person ‘in a flash’ what the optimal way is to feel. Recall here Aristotle's concession that even the virtuous will waver and falter at times, and that an emotional virtue such as mildness (the affective golden mean with respect to anger) may even require its own intermittent sub-optimality for it to continue to be a virtue.Footnote ⁵⁷ I am afraid that these concerns render Wolf's all-or-nothing conception of phronesis stillborn.

We need to look elsewhere for guidance. Before doing so, however, let me enter the same caveat that Wolf does: I am not aiming at the correct textual exegesis of Aristotle but at a possible way around the difficulty that the moral problem of emotional ambivalence creates for an Aristotelian type of virtue ethics. Where better to look than in the abundant social-scientific research on bicultural integration. Biculturals have to negotiate the demands of conflicting emotions and virtues not only intermittently, like most of us, but rather all the time – or at least all the time they are exposed concurrently to the two cultures. How do successful biculturals cope? Until recently, the emphasis in the social-scientific literature was on integration qua ‘hyphenated identity’ or ‘compartmentalisation’. Individuals who adopt this strategy successfully act not as unified cultural beings but as unions of different cultural beings, diverging and converging to the needs of the moment by engaging in cued cultural frame-switching. Even if such compartmentalisation helps solve existential problems of bilculturalism, however, it offers little reprieve for our moral problem. The second paradigm, which is gradually emerging from research on bicultural integration, could be termed integration qua ‘synergic identity’. Synergic integrators have forged an entirely new, fused identity, a unique configuration which cannot be reduced to its parts. When biculturals negotiate identity in this truly integrated way, they have managed to turn their cultural traditions into objects of higher-order reflection by cognitively juxtaposing them; and they consider their double background not as a handicap but rather as a unique source of interpretative tools for grasping experience. There is no more code-switching, but rather full synthesis, characterised by a sense of self-fulfilment.Footnote ⁵⁸

Experiences of synergic bicultural integration are typically described as those of an ‘inner voice’ or a ‘third ear’ guiding the self towards a unified mode of acting and being. Inevitably, phronesis comes to mind as a comparison. Feeding on emotional dispositions cultivated unreflectively in the young moral agent through habituation, phronesis – after it comes into play – re-evaluates those dispositions critically, allowing them to ‘share in reason’. The phronesis ‘concerned with the individual himself seems most of all to be counted as [phronesis]’, which indicates that phronesis is most concerned with critical self-cultivation. Phronesis also compares the relative weight of competing values and emotions – values and emotions that are incompatible but not incommensurable – with eudaimonia: the ultimate good and unconditional end of human beings. For phronesis enables one to ‘deliberate finely’ not only about what is good for oneself in ‘some restricted area’ but about ‘what promotes living well in general’. This involves reasoning, based on ethical first principles, about one's appropriate and rational combinations of desires and beliefs. Therefore, a person who has acquired phronesis has – in the Aristotelian model presupposed here – the wisdom to adjudicate the relative weight of different virtues in conflict-situations and to reach a measured prescriptive verdict about what to feel and do.Footnote ⁵⁹

This mediating, overseeing and ‘prescriptive’ role of phronesis gives it a clear status as a second-order intellectual virtue.Footnote ⁶⁰ ‘Intellectual’ must be understood broadly, however, as phronesis incorporates not only true beliefs about what to feel and do, but also includes a motivational component: the desire to adhere to those true beliefs: a desire whose satisfaction or frustration, in turn, supplies the affective component of the emotion.Footnote ⁶¹Phronesis operates here, then, very much like a second-order meta-emotion (a cognition–desire combination directed at other emotions and dispositions). Indeed, it seems to fit perfectly current characterisations of such emotions:

Phronesis first moulds and forms the person's emotional dispositions, in order to turn them into emotional virtues; then it stands guard and comes to the rescue again if two emotional virtues conflict in a given situation, as in Case 7. It does not suppress the non-optimal emotion in such cases – the emotion will still be felt (see assumption (b) in Section 3) – but it allows the non-optimal emotion to be compared with the optimal emotion from the general standpoint of the agent's eudaimonia and infuses the agent's deliberative process with reason once again so that the agent realises what is the proper way to feel – and becomes motivated to feel that way.

5. Concluding Remarks

If the above characterisation of the mediating role of phronesis is correct, where does that leave ‘The Mystery of the Missing Motivation’, and Carr's and Stark's conflicting responses to it?

Although Carr clearly commits certain missteps in his criticism of Stark, he is right overall that the key to the solution of the moral problem of emotional ambivalence lies in assumption (e). The claim that virtuous agents are motivationally unified does stand in need of serious qualification. The non-optimal emotion in Case 7, and in other similar cases, continues to exist, reminding us of the dilemmas that we face: The meta-emotion of phronesis, however, allows the virtuous person's first-order emotion to ‘share in reason’ again and be outweighed by reason (which is not the same as being suppressed by reason). If there is ‘motivational unity’ here, it is very much one of an intense dialectical synthesis. Stark's motivational externalism fails because all emotions motivate, however weakly: simply put, they do not evanesce, or remain mere ‘normative reasons’, and assumption (c) cannot be dropped. Nevertheless, the account that I have offered retains some of the externalist flavour of Stark's solution in that the desire to feel happy for my friend in Case 7 is (given that I am a virtuous agent) ultimately external to the original emotion of happiness itself, emanating rather from the meta-emotion of phronesis.

Carr has emphatically rejected therapeutic accounts of emotional regulation, such as ‘Emotional Intelligence’, for their insistence on ‘complete emotional harmony’ and unproblematic resolution of ‘unease and conflict’. He considers such endeavours both psychologically disabling and morally untoward. In contrast, he argues, ‘emotional ambivalence, conflict and disquiet, even at the price of some practical dysfunctionality, cannot but be part and parcel of any recognisable human condition’. On his alternative model of conflictual unity, the virtuous are ‘precisely those equipped with the richly complex – albeit conflicted – psychological life through which alternative possibilities of (virtuous) action remain available’.Footnote ⁶³ There is no incompatibility between this view and that of Pugmire who emphasises the extent to which emotionally ambivalent agents will be susceptible to psychological dysfunction (recall Section 2). After all, in the Aristotelian model, only a small minority of adults are virtuous agents; the majority persist somewhere between the levels of incontinence and continence.Footnote ⁶⁴ Only for the virtuous – and for committed ‘upwardly-mobile’ moral learners at lower levels – will emotional ambivalence work as a source of further sensitisation and character-perfection.

There are at least two ways in which we can envisage emotional ambivalence to be character-building. One is through the exploration of alternative possibilities: Very much like Mill's argument about the need to have truth constantly challenged in order for it to retain its heartfelt vitality, one could argue that optimal emotions would lose their urgency and immediacy – and ultimately their motivational bite – if they did not regularly come into conflict with other (non-optimal) emotions. But one could also argue – and this would make something of a virtue out of Pugmire's earlier depiction of ambivalent emotions as ‘inchoate’ and ‘unconsolidated’ – that the optimal emotion will, in scenarios such as Case 7, be tempered by, or contain residues of, the non-optimal one even after the phronesis-inspired arbitration process, and will thus reflect more adequately the complexity of a mature and well-rounded emotional life.Footnote ⁶⁵ One must avoid, however, going too far in fetishising the tragic if one is to remain faithful to an Aristotle-inspired virtue ethics.Footnote ⁶⁶ For most people, mixed emotions are a mixed blessing, if not a curse, and only for a chosen few will the label ‘constructive dividedness of mind’ be apt.