(PDF) Doing without emotions

This is the accepted manuscript of a paper published in Pacific Philosophical Quarterly 93(3) on 9th July 2012. The definitive version is available at http://onlinelibrary.wiley.com/doi/10.1111/j.1468-0114.2012.01428.x/pdf 1 Doing without emotions Carolyn Price Abstract This paper considers a central question in the philosophy of emotion: what is an (instance of) emotion? This is a highly controversial question, which has attracted numerous answers. I argue that a good answer to this question may prove very hard to find. The difficulty, I suggest, can be traced back to three features of emotional phenomena: their diversity, their complexity and their coherence. I end by suggesting that we should not be disturbed by this result, as we do not need to know what an instance of emotion is in order to investigate the topic of emotion. 0. Introduction The philosophical literature on emotion is littered with attempts to say what an emotion is. In this paper, I too am going to investigate this question. My aim, however, is not to propose an answer. Instead, I shall argue that, on one interpretation, the question may be impossible to settle. This need not worry us, I shall suggest, because there is no pressing need to settle it. The qualification ‘on one interpretation’ is important, because the question ‘What is an emotion?’ can be understood in at least three different ways: Q1 The word ‘emotion’ can be used to name a broad class of psychological phenomena, which includes anger, sorrow, and joy. In asking ‘What is an emotion?’, we might be asking why anger, sorrow and joy should be classified as emotions, and not (say) as desires, moods or traits. Q2 The word ‘emotion’ can be used to refer to a type of emotion, such as anger, sorrow or joy. In asking ‘What is an emotion?’, we might be asking how one type of emotion differs from another. 2 Q3 The word ‘emotion’ is sometimes used to refer to a particular psychological occurrence. We might say, for example, ‘Her emotion was short-lived’ or ‘His emotion clouded his judgement.’ In asking ‘What is an emotion?’, we might be asking what these occurrences are: what is it that lasts only a short time, or that clouded his judgement? Q1 and Q2 concern types of emotion: they are questions about classification. Q3, in contrast, is concerned with instances of these types: it is a question about individuation. My argument is concerned only with Q3. This might seem odd. Surely, we might think, Q2 and Q3 must be answered together: if instances of emotion are intentional states, they must be classified by intentional content; conversely, if they are classified by phenomenal character, they must be feelings, and so on. However, this is not the case. This is because in answering Q2, we might look, not to instances of emotion themselves, but to their causes, their effects, or the psychological context in which they occur. It would be perfectly possible, for example, to combine the view that instances of emotion are feelings with the view that they should be classified according to the intentional properties of the perceptions that cause them (Schaffer, 1983: 161-2). Q2 and Q3, then, are independent questions. What counts as having a good answer to Q3? First, I shall assume that we are not looking for a stipulative answer, but for one that is backed by some reasonable principle. Secondly, I shall assume that this principle must be decisive, in that it excludes other possible answers to the question. Thirdly, I shall assume that we are looking for an answer that is theoretically interesting – one to which we might usefully appeal in building a theory of emotion. There is no shortage of possibilities: instances of emotion have been identified with bodily feelings (James, 1890); evaluative judgements (Lyons, 1980; Solomon, 1993; Nussbaum, 2001); thoughts or perceptions (Roberts, 1983; Stocker, 1987); 3 appraisals or evaluations of a distinctive kind (Helm, 2001; Prinz, 2004); complexes of desire and belief (Marks, 1982); brief, automatic, programmed sequences of physiological and psychological changes (Ekman, 1992; Delancey, 2002); enduring and evolving psychological processes, involving feelings, desires and thoughts (Goldie, 2000). To some extent, this controversy is explained by disagreements about the nature of specific emotional phenomena. For example, Roberts, Solomon and Helm agree that instances of emotion are emotional evaluations; they disagree only about what emotional evaluations are. But even when disputes of this kind are put to one side, Q3 remains controversial. In what follows, I shall argue that this controversy may prove very difficult to resolve. The difficulty arises, I shall suggest, from three features of emotional phenomena: their diversity, their complexity, and their coherence. My argument, then, assumes a certain picture of emotion. Although I believe that it is a plausible picture, I accept that it is open to question. For this reason, my conclusion is a conditional one: if emotional phenomena are diverse, complex and coherent in the ways that I shall suggest, a good answer to Q3 may prove very hard to find. Moreover, my conclusion is a provisional one. This is because I shall proceed by considering some different strategies that might be used to answer Q3. There is always the possibility that there is some further strategy that I have overlooked. My aim, then, is not to hammer my conclusion home, but to raise a possibility, and to suggest another way forward. The paper is divided into four sections. In the first section, I lay out the groundwork for my argument by setting out the picture of emotional phenomena that I shall assume in what follows, and I go on to introduce the three key features that I take to characterise emotional phenomena – diversity, complexity and coherence. Of these three claims, the coherence claim is the most controversial, and I shall spend a little time defending it. I begin the second section by distinguishing two types of answer to 4 Q3: single component theories and complex process theories. The remainder of this second section is devoted to single component theories. I consider four different strategies that might be used to defend a single component theory against rival accounts, and I argue that none of them succeed. I concede, though, that my argument against one of these strategies (the explanatory strategy) rests on the most controversial element in my account of emotional phenomena – the coherence claim. I shall not claim, then, that that this strategy certainly fails, but only that there is reason to doubt it. In the third section, I turn my attention to complex process theories. I argue that the diversity of emotional phenomena makes it impossible to make a principled choice between different kinds of complex process theory. I shall briefly consider, too, the possibility that we might adopt a pluralistic approach. In the fourth section, I end with a (very brief) suggestion as to how we might manage without an answer to Q3. My aim is not to insist upon this particular response to the problem, but simply to establish that there is room for an alternative approach. 1. Laying the groundwork 1.1. Emotional phenomena I shall begin by setting out the picture of emotional phenomena on which my argument depends. In doing this, I am not attempting to answer Q3: the term ‘emotional phenomenon’, as I shall use it here, is not a synonym for ‘instance of emotion’. Rather, it is meant to refer to any psychological or physiological event or state that can be described as ‘emotional’: these might include psychological episodes, thoughts, feelings, motivations, and so on. Hence, emotional phenomena might include not only instances of emotion, but also their causes, effects, components and concomitants. In describing these phenomena, I shall paint with a very broad brush. This is because I am trying to avoid controversies that are irrelevant to my argument. 5 Suppose that Alice is angry with Bill. This claim can be understood in several ways. One possibility is that Alice is bristling with anger at some thoughtless act. Bristling with anger is an example of what I shall call an emotional reaction. Emotional reactions are relatively brief, lasting perhaps a few seconds, and they seem to be involuntary. They involve a number of ingredients: when Alice reacts angrily to Bill’s behaviour, she perceives his behaviour as an offence; her attention switches to him; her muscles tense; she frowns; she has an urge to retaliate; she feels angry. Alice’s initial reaction may develop into a full-blown emotional episode. As the episode develops, the physiological and psychological changes that constituted her initial reaction will continue. In addition, her angry reaction is likely to unleash a flood of angry thoughts about Bill and his behaviour, thoughts or fantasies about what to do, and further emotional reactions. If her anger is not too intense, Alice might try to control it – by counting to ten, say. An emotional episode may be quite brief, lasting a few minutes. Arguably though, emotional episodes can extend over much longer periods – hours, days, or even weeks: Alice might still be angrily complaining about Bill’s behaviour a fortnight later. Some theorists of emotion reject this way of characterising the situation. What we have here, they suggest, are separate episodes of anger, linked by a disposition to become angry whenever Alice recollects Bill’s behaviour.1 Indeed, in many cases, this will be the right thing to say. In some cases, though, enduring anger involves more than a disposition to become angry again: it involves an active propensity to think about the offensive incident, to raise it in conversation, and to interpret further events in its light. In these cases, it is natural to regard the subject’s angry outbursts as symptoms of a continuing process, which may influence the subject’s thoughts and motivations for some time. Finally, some emotional states that have the potential to last a lifetime. These include emotional dispositions, such as fear of snakes. But not all enduring emotional 6 states are wholly dispositional. To say that Alice harbours an angry attitude towards Bill is not to say just that she is disposed to be angered by him: it implies that she has an active propensity to think about his behaviour towards her, and a continuing sense that his treatment of her is unfair. Similarly, perhaps many people are disposed to become jealous should they realise that their partner has been unfaithful; but to harbour a jealous attitude implies more than this: it implies an active propensity to look out for signs of infidelity and a continuing sense that the relationship is under threat. I shall refer to states of this kind as emotional attitudes. Like the other emotional responses that I have described, an emotional attitude will be underpinned by an evaluation of its object; it is likely to involve certain motivations and emotional dispositions, and an active propensity to pay attention to its object.2 1.2 Diversity, Complexity and Coherence The boundaries between emotional reactions, episodes and attitudes are not clear-cut: reactions may shade into episodes, episodes into attitudes, and attitudes into dispositions. But there are some significant differences between them. For example: 1. Emotional reactions seem to be involuntary, yet people can sometimes control emotional episodes and attitudes. 2. Emotional reactions leave little space for thought; in contrast, emotional episodes and attitude can involve thought in several ways – ruminating about the situation, searching for a solution, fantasising a response, and so on. 3. Emotional reactions and episodes are relatively transient events, which need leave no lasting trace on the subject; emotional attitudes, in contrast, are likely to have a significant influence on the subject’s values and outlook. Hence, when we raise philosophical questions about emotion – questions about the role of emotional capacities in our ethical lives, or questions about the relationships between 7 emotion, reason and autonomy – it may well matter what kind of emotional response we have in mind. In this sense, emotional reactions, episodes and attitudes constitute interestingly diverse kinds of emotional response. On the other hand, emotional reactions, episodes and attitudes share some significant features. One obvious similarity is that they are all complex responses, involving a combination of evaluation, motivation, feeling, and so on. Conversely, emotional evaluations, motivations and feelings do not characteristically occur in isolation, but as components of complex processes or states. Indeed, it might be suggested that complexity is an essential feature of emotion – that an aggressive thought occurring in isolation would not constitute an angry thought. I shall refer to this possibility later on; but my argument depends only on the claim that emotional responses are characteristically complex. I have made two claims about emotional phenomena: (1) Diversity: there are theoretically interesting differences between emotional reactions, episodes and attitudes. (2) Complexity: emotional responses are characteristically complex. My third claim – the coherence claim – will take a little more explaining. It can be summarised as follows: (3) Coherence: generally speaking, emotional responses of a given type share a recognisable structure, a structure that can be explained, to a significant degree, in functional terms. Complex emotional responses, I take it, are not random collections of symptoms: typically, they involve a recognisable pattern of cognitive, motivational and behavioural changes, characteristic of the type of emotion involved. In her anger, for example, Alice evaluates Bill’s behaviour as an offense; her attention is focused on Bill; she is motivated to retaliate; and so on. Angry reactions, episodes and attitudes involve 8 the same basic pattern of evaluation, attention and motivation. This is not to say that the shape of an angry response can be predicted in detail, or that angry responses always develop in exactly the same way, regardless of the situation. My claim is just that there are some basic ingredients that angry responses characteristically share. Moreover, anger is naturally viewed as a coherent, coordinated response to the situation. The effect of an angry response is to marshal a range of physiological and psychological resources in a way that prepares and motivates the subject to retaliate – by attacking, threatening or scolding the offender. Retaliation makes sense as a response to offence in that it can cause the offender to back down, so deterring further offence. I am not claiming that responding with anger is the best way to achieve this goal, or even that it is usually effective. My claim is that an angry response appears to be designed to deal with offence. Moreover, the same point can be made about many other kinds of emotion: fear, guilt, shame, gratitude, envy, jealousy, contempt – among others – can all be seen as coherent, coordinated responses to the situations that elicit them. It is tempting to suppose that this is no accident. Plausibly, emotional responses are supposed to have these effects: this is their function or purpose. This functional claim, I take it, rests on an appeal to history: it implies that angry responses take the form that they do because, in the past, similar responses have sometimes been effective in deterring offence.3 How this should be filled out will depend on whether our emotional capacities have been shaped primarily by evolutionary history, by social and cultural influences, by personal experience, or by some combination of these.4 As far as my argument goes, it does not matter which of these stories is true: what matters is the claim that the structure of an angry response can, to a significant degree, be explained in functional terms; and that this is true of many other types of emotion too. Of the three claims that I have made in this section, the coherence claim is the most controversial. In particular, the claim that the structure of an emotional response 9 can be functionally explained raises a number of questions. For this reason, I shall end this section by saying a little more about this claim, and by offering one consideration in its favour. It is important to note that the coherence claim is qualified in a number of ways. First, nothing I have said here excludes the possibility that the processes that shape our emotional capacities might occasionally give rise to functionless emotions.5 Again, the claim does not imply that every component of an emotional response has a function. Nor does it imply that the fine details of a particular emotional response can be functionally explained: these are likely to depend on the situation and on the subject’s other psychological states. The claim applies only to the broad structure of the response. Even so, it might be objected that there are other ways in which we might explain why emotional responses are structured as they are. Consider, in particular, the case of emotional motivation. I have suggested that an angry response characteristically involves an urge to retaliate – perhaps by attacking, threatening or scolding the offender. According to the coherence claim, the explanation for this lies in the past: anger motivates this kind of behaviour because, in the past, this kind of behaviour has sometimes been an effective response to offence. But there are other possibilities. One possibility is that the explanation lies with people’s values and beliefs: in most cases, perhaps, people want to deter offensive behaviour and believe that attacking, threatening or scolding the offender are effective ways to do this. Another possibility is that the explanation has to do with the nature of the situations in which anger is characteristically elicited. Jesse Prinz has made a suggestion of this kind. Prinz denies that emotions motivate specific kinds of behaviour, suggesting instead that negative emotions, such as anger, simply motivate the subject to bring the situation to an end. As it happens, attacking, threatening or scolding the offender are generally effective ways 10 to deal with a situation of offence. This is why angry people can usually be expected to behave in these ways (Prinz, 2004, 173-4). There is a problem with these alternative suggestions, however. The problem is that emotional motivations are sometimes out of step with the situation, and at odds with the subject’s beliefs about what to do. In her anger, Alice may be strongly motivated to scold Bill, even though doing so is evidently counterproductive, and even though she realises that this is the case. This would be puzzling, if Alice’s motivations depended on her beliefs or on the details of the situation. But if the coherence claim is correct, there is no puzzle: the fact that retaliation has sometimes proved effective in the past is no guarantee that it will be effective on any particular occasion. One point in favour of the coherence claim, then, is that it can explain the intelligibility of emotional responses overall, while allowing for their recalcitrance on particular occasions. This hardly constitutes a full defence of the coherence claim. To offer such a defence, it would be necessary to consider a much wider range of emotions and emotional phenomena, and to do so much more comprehensively than I can do here. Moreover, even if the claim is correct, there is plenty of scope to investigate which aspects of an emotional response are functionally explicable and which would be better explained in other ways. My aim here has not been to establish that the claim is true, but to establish that the claim is worth taking seriously. 2. Single Component Theories 2.1. Complex processes or single components? In section 1.1, I introduced a number of emotional phenomena. These included complex processes, such as emotional reactions, episodes and attitudes, and the components of such processes, such as emotional feelings or evaluations. Which of these should we take to be an instance of emotion? 11 Some theorists suggest that instances of emotion should be identified with complex emotional responses. Paul Ekman (1992) can be read as identifying instances of emotion with emotional reactions, or perhaps brief emotional episodes. In contrast, Peter Goldie (2000: 11) identifies instances of emotion with more extended emotional processes – emotional attitudes, perhaps. We might call these views complex process theories. Other theorists have identified instances of emotion with components of these processes, for example, feelings or evaluations. We might call these views single component theories. There are also dual component theories: for example, Marks’ (1982) suggestion that an instance of emotion is a belief coupled with a desire. How might we decide between these suggestions? First, we need a description of what emotional phenomena are actually like – a more detailed version, perhaps, of the description offered in Section 1. In developing this description, we might draw on commonsense experience, on empirical research, or both. In what follows, I shall assume that it is possible in principle to draw on these resources to achieve an accurate, illuminating and agreed description of emotional phenomena. Secondly, we need a principle that will take us from this description to a claim about what an instance of emotion is. As we have seen, the principle to which we appeal should be a reasonable one, and it will need to generate a decisive and theoretically interesting answer to Q3. I shall consider four principles to which a single component theorist might appeal to defend their account. I shall argue that none of them succeed, and I shall try to explain why. 2.2. The matching strategy Perhaps the simplest strategy that a single component theorist might take is the matching strategy. The strategy proceeds in two stages: the first step is to assemble some uncontroversial description of what instances of emotion are like; the second is to 12 find the component that matches this description. One particularly clear example of this approach is Roberts’ (1983) argument for the claim that instances of emotion are emotional evaluations (understood as ‘concerned construals’). Roberts begins by assembling a list of properties that, he suggests, characterise instances of emotion (Roberts, 1983: 183-4). What follows is a slightly simplified version of Roberts’ list: 1. In paradigm cases, instances of emotion are felt. 2. They are intentional states. 3. They are typically accompanied by physiological changes, some of which are felt. 4. They typically depend on, but sometimes conflict with, our beliefs. 5. Some types of emotion imply a disposition to behave in certain ways. 6. They are sometimes, but not always, subject to voluntary control. 7. They are typically experienced as unified states of mind, rather than as sets of components. Roberts goes on to argue that emotional evaluations, as he understands them, have just these properties. He concludes that instances of emotion are emotional evaluations. One objection that might be made to this argument is that it does not provide a decisive answer to Q3. This is because there are other answers that are compatible with Roberts’ list. On the face of it, for example, the list is compatible with the claim that instances of emotion are desires or the claim that they are emotional episodes.6 Still, this objection does not undermine the matching strategy. This is because the list might be expanded or refined in a way that decides between rival theories. For example, perhaps the suggestion that instances of emotion are desires would be ruled out by a more detailed account of their intentional properties. However, the matching strategy faces another, less tractable objection. I have granted that it is possible to develop an accurate and agreed description of what 13 emotional phenomena are like. But the matching strategist needs more than this: they need to begin from an accurate and agreed description of what instances of emotion are like. That is a much tougher demand. Read in this way, Roberts’ list incorporates some controversial assumptions. For example, he suggests that instances of emotion are themselves intentional states, but that they are merely accompanied by bodily feelings (Roberts, 1983: 207). In contrast, a proponent of the view that instances of emotion are bodily feelings will insist that instances of emotions themselves have a bodily feel, while holding that they are merely accompanied by intentional states. Meanwhile, a complex process theorist might insist that these are both properties of the emotion itself. If the matching strategy is to succeed, we must be able to find an agreed description of instances of emotion that can decide between these claims; and we need to do this without presupposing an answer to Q3.7 Empirical research might provide us with a more accurate and detailed description of the properties of emotional phenomena. But this will not, by itself, tell us which of these properties belong to the emotion itself. More puzzlingly, common sense does not seem to yield an answer either. All the positions described in the last paragraph are compatible with commonsense assumptions about emotion: none of them involve denying, for example, that being angry involves perceiving the situation in a certain way, feeling tense and agitated, and so on. How have we reached this impasse? The problem, I suggest, has to do with the complexity of emotion. Earlier, I suggested that emotional evaluations, feelings and motivations are not characteristically experienced in isolation, but rather come as a package. As a result, it is possible to have a perfectly adequate conception of what is involved in being angry without needing to decide whether intentionality – or controllability, or a certain phenomenal character – is a property of the anger itself, or of some other phenomenon that is intimately and reliably linked to it. If this is correct, we 14 cannot expect to begin with a clear and agreed conception of what instances of emotion are like. To develop such a conception, we already need an answer to Q3. 2.3. The causal strategy Perhaps this is too quick. For, it might be suggested, there is one context in which commonsense psychologists do need to distinguish an instance of emotion from the phenomena that characteristically accompany it. This is in appealing to people’s emotional responses to explain their behaviour, thoughts and so on. Consider, for example, the following commonsense explanation: S1 Alice stalked out of the library because of her anger. On the face of it, this explanation identifies Alice’s anger as the cause of her behaviour. If so, then, as far as common sense goes, Alice’s anger and her angry behaviour must be two different things. Assuming that there is no reason to challenge common sense on this point, we may have found a way to distinguish an instance of emotion from other phenomena that are its causes and effects. I am not aware of any theorist who has explicitly appealed to this strategy. Nevertheless, it seems sufficiently tempting to be worth considering, if only to note that it will not work. To see the difficulty, consider the following explanations: S2 Because of her anger, Alice assumed that Bill meant to provoke her. S4 Alice wanted to slap Bill because of her anger S4 Alice felt tense and jumpy because of her anger S5 Alice’s heart was pounding because of her anger. S6 Alice saw Bill’s behaviour as offensive: her anger was because of that. Like S1, these all appear to be perfectly acceptable, commonsense explanations. But if so, it poses a serious problem for the causal strategist. For it looks as if the strategy rules out all single component theories: Alice’s anger cannot be identified with her 15 evaluation, her thoughts, her urges, her feelings, or the physiological changes she undergoes. It looks as if nothing is left to constitute Alice’s anger. What, then, could Alice’s anger be? Perhaps it is some further component of her angry episode that we have yet to identify. But it is hard to see what this could be. A more plausible possibility is that at least one of these explanations is not a causal explanation, but a constitutive one. But which one? This question would be easy to settle if we already had an answer to Q3; but without one, there seems to be no way in which to decide the issue. Certainly, the explanations themselves offer no obvious clue. In this case, there is more than one way to diagnose the problem. One possibility is that the explanations listed above do indeed include a mix of causal and constitutive explanations. But even if this is so, there may be no pressing need for us, as commonsense psychologists, to decide which is which. The explanations work because anger is reliably correlated with having certain kinds of thoughts, feelings and so on: whether the correlation is causal or constitutive need not be important. In other cases (balls breaking windows, for example) it is important to make this distinction: this is because, when the correlation is causal rather than constitutive, it is possible to intervene between cause and effect. In the case of emotion, however, there is ordinarily no such possibility: controlling one’s angry thoughts, motivations or feelings is not a matter of intervening between one’s anger and its effects, but of controlling the response as a whole.8 And this is because angry thoughts, feelings and so on characteristically come as a complex package. Another possibility is that that the causal strategy has gone wrong from the start: none of the explanations listed above are causal explanations.9 Rather, they are all constitutive explanations, which work by identifying Alice’s thoughts, desires and so on as components of a recognisable pattern of psychological response (Kenny, 1963; Goldie, 2000: 42-43). We can understand these explanations without deciding which 16 component of the process (if any) constitutes Alice’s anger: we just need to know what is characteristically involved in being angry. This diagnosis appeals, not to the complexity of emotion, but to its coherence – specifically, to the idea that emotional responses have a recognisable structure, to which we appeal in explaining why Alice responds as she does.10 A single component theorist might favour the first of these diagnoses; a complex process theorist will favour the second. Whichever diagnosis is correct, the causal strategy is not well-placed to deliver an answer to Q3. 2.4. Two essentialist strategies An emotional response, as we have seen, has many components. Nevertheless, it is sometimes suggested that one of these component is an essential ingredient in an emotional response, while other components are inessential. Theorists have sometimes appealed to this idea in the search for an answer to Q3. In fact, there are two different ways in which this strategy can be used: it can be used negatively – to exclude a particular answer to Q3; and it can be used positively – to support a particular answer. The negative strategy is commonly used to attack William James’s claim that instances of emotion are bodily feelings. James’ opponents object that there are paradigm cases of emotion that do not involve bodily feelings. If they are right about this, this does indeed constitute a good reason to reject the view that instances of emotion just are bodily feelings (Solomon, 1993: 99; Helm, 2002: 34-5). It is possible to envisage other variations on the theme. For example, Roberts (1983: 189-90) claims that emotional responses need not be conscious; he concludes that instances of emotion cannot be feelings of any kind. Similarly, it might be argued that that there are emotional responses – cases involving emotional contagion, say – that do not involve evaluation; if so, this would suggest that instances of emotion are not evaluations. 17 Could this negative strategy be used to find a decisive answer to Q3 by a process of elimination? There are two obstacles to success. First, it is far from clear that this strategy can be carried through without appealing to non-paradigm or controversial cases. Arguably, all paradigm cases of emotion involve a mix of feeling, evaluation and perhaps changes to attention and motivation too. Indeed Roberts himself allows that unconscious emotional responses are not paradigm cases (Roberts, 1983: 189); responses triggered by emotional contagion are not paradigm cases either. But this leaves the door open to opposing theorists simply to deny that these are genuine cases of emotion at all. Secondly, this negative strategy cannot be used to exclude a complex process theory: for complex process theorists need not accept that emotional responses have any essential ingredients at all. However, neither of these objections undermines the positive strategy. A proponent of this strategy does not need to assume in advance that emotional responses have some essential ingredient. Rather, they need to begin by establishing that there is, in fact, such an ingredient. Nor do they need to confine their appeal to paradigm cases: they might be able to show that this ingredient is shared by all possible cases of emotion – paradigm or otherwise. If such an ingredient can be found, it might well be argued that it would be reasonable to take it to be an instance of emotion: for identifying this ingredient, we have identified what is essential to emotion. Hence William James identifies instances of emotion with bodily feelings on the grounds that bodily feelings are essential to emotion (James, 1890: 451-2). Bennett Helm might be read as employing a similar strategy when he claims that felt evaluations are what emotions ‘essentially are’ (Helm, 2002: 34). There are at least two difficulties with this approach. First, even if we could identify such an ingredient, it remains possible that this is not all that is essential to emotion. Emotional responses might have essential structural or functional properties 18 too. Hence, to identify an instance of emotion with a single ingredient risks overlooking another, equally important, aspect of emotional experience. Of course, it is no more obvious that emotional responses must have a particular structure than that they must include a particular component. My point is just that there is no quick move from the claim that emotional responses share some essential ingredient to the claim that, in identifying this ingredient, we have captured all that is essential to emotion. Hence, even if emotional responses do contain some essential ingredient, the positive essentialist strategy may fail to yield a decisive answer to Q3. More importantly, the principle underlying this positive version of the strategy looks weak. To see this, consider an analogy. Suppose that people start falling victim to a new disease, known as birdpox. Bouts of birdpox can involve a range of symptoms: a fever, swollen joints, a rash, and so on; one of these symptoms – the rash – is invariably present. People soon come to have a clear idea of what a bout of birdpox involves. But what, they ask, is birdpox itself? There are two ways in which we might reasonably answer this question. We might identify birdpox with a syndrome: the symptoms that constitute a bout of the disease. Or we might identify birdpox with whatever causes the symptoms – a viral infection, perhaps. But there is no pressing reason to identify birdpox with the rash, essential or not. The essentialist might protest that this is unfair. If bouts of birdpox are always produced by a viral infection, the infection will also be an essential ingredient of a bout of birdpox. And I have, after all, conceded that it might be plausible to identify birdpox with the infection. However, this is not enough to counter the objection: it suggests that the essentialist should identify birdpox with the combination of infection and rash; but this seems just as implausible. Our readiness to identify birdpox with the infection seems to rest on a different principle. In the next section, I shall suggest what it is. 19 2.4. The explanatory strategy I will end this part of the paper by considering what I take to be to be the most promising strategy for the single component theorist. This is the approach taken by Prinz (2004). Identifying an instance of emotion with a single component, Prinz suggests ‘is a way of drawing attention to the feature that is most fundamental to understanding emotions’ (Prinz, 2004:18). I take it that what Prinz means is this: suppose that we can find some component (or combination of components) within a complex emotional response that plays a key role in explaining how the response develops. If so, we would have a good reason to take this component to be the emotion. Arguably this is precisely what we are doing when we identify birdpox with the viral infection: what makes the infection a good candidate to be birdpox is that it plays a fundamental role in explaining why a bout of birdpox develops as it does. Once again, the success of the explanatory strategy turns on a certain empirical assumption – the assumption that there is a component that plays this key explanatory role. As before, I shall concede that this might be the case. Prinz holds that there are two items that fill this role (Prinz, 2004: 55-67; 163). The first is what he calls ‘an embodied emotional appraisal’: on Prinz’s account, this is both a bodily feeling and a signal that some emotionally salient situation has arisen. The second is a ‘valence marker’, which tags the situation as good or bad. According to Prinz, then, it is the content of Alice’s embodied appraised, together with the value – positive or negative – of the valence marker which explains why Alice’s emotional response takes the form that it does. The details of Prinz’s account are important and interesting. They are also controversial.11 Nevertheless, theorists who reject his account of what an emotional evaluation is might well agree that emotional evaluations play a fundamental role in explaining how an emotional response develops, and that this constitutes a good reason to identify instances of emotion with emotional evaluations. 20 As we have seen, however, there is another way in which we might explain why Alice’s angry response takes the form that it does. This is to appeal to the idea that anger is a coherent response, designed to perform a specific function – that is, to prepare and motivate the subject to deal with offence. This functional explanation involves looking at Alice’s angry response as a whole to understand how its components work together to produce a particular outcome. Hence the availability of this kind of explanation might be taken to support a complex process theory. Moreover, this top- down, functional explanation is not a rival to the causal explanation suggested by Prinz. Both explanations might be true, because they are explanations of different kinds.12 If so, the explanatory strategy will generate two different answers to Q3. Prinz is aware of this line of response. He replies by comparing a complex process theorist of emotion to someone who insists on giving a complex process theory of visual experiences, such as seeing something red. When someone sees something red, Prinz allows, they characteristically undergo other psychological changes – for example, they may attend to the red thing, form certain motivations regarding it, and so on. Hence, visual experiences are not characteristically had in isolation but within a broader psychological context. Nevertheless, as Prinz points out, there is no pressure to insist that we should cease to regard visual experiences as simple psychological entities, and instead view them as complex processes, including what we would otherwise think of as their cognitive and motivational effects (Prinz, 2004: 242). Given the picture of emotion I have adopted here, however, there is a disanalogy between the two cases. Visual experiences characteristically occur within a wider psychological response, but they do not characteristically occur within a psychological response of a specific kind. We might see something red without forming the belief that there is something red there; and while a visual experience may give rise to a motivation of some kind, it does not give rise to a motivation of a particular kind. The point of such 21 an experience is to register the information that there is something red there; but this information need not be taken up by other systems; and when, it is, it may be taken up in different ways. The many contexts in which such experiences may occur makes it worth picking them out as discrete states. In contrast, I have suggested, an emotional evaluation characteristically occurs in the context of a specific kind of psychological response – a complex emotional response; and in that context, it has a specific role to play. Hence, picking on emotional evaluations as the fundamental emotional phenomenon draws our attention away from an equally fundamental aspect of emotion: that emotional phenomena are or form part of psychological processes that characteristically have a particular kind of structure and function. If so, the explanatory strategy fails to yield a decisive answer to Q3. It should be clear that my argument against Prinz depends on a substantive claim about the nature of emotional responses – that is, the coherence claim. If the coherence claim is true, the explanatory strategy will fail because there are two, mutually compatible ways of explaining why an emotional response takes the form that it does. Earlier, I argued that the coherence claim deserves to be taken seriously. If so, we should take seriously the possibility that the explanatory strategy fails to yield a decisive answer to Q3. 3. Complex Process Theories 3.1 Reactions, episodes or attitudes? In the last section, I considered four different strategies that might be used to support a single component theory. I argued that, if emotional responses are complex and coherent in the ways that I have suggested, none of these strategies are likely to succeed. It might be suggested, then, that we should jump the other way, and adopt a complex process theory of emotion. Certainly, complex process theories have some 22 advantages: in particular, they can be viewed as reflecting the complexity of emotional responses and the importance of their structural and functional properties. Nonetheless, it is worth bearing in mind that nothing I have said so far implies that we should not adopt a single component theory, only that there is no decisive reason to prefer a single component theory to a complex process theory. Moreover, complex process theories run into trouble once we take into account a further feature of emotional responses: their diversity. The essence of the problem is that there are several kinds of complex emotional response: emotional reactions, emotional episodes and emotional attitudes. Which of these constitutes an instance of emotion? As we have seen, there is room for disagreement about this: Paul Ekman (1992) takes an instance of emotion to be (roughly) an emotional reaction; whereas Peter Goldie (2000: 11) takes it to be (roughly) an emotional attitude. How could we decide between these suggestions? Ekman supports his account by appealing to a close cousin of the essentialist strategy: he suggests that identifying instances of emotion with brief emotional reactions captures everything that is distinctive of emotional phenomena (Ekman, 1992: 195). In contrast, emotional attitudes might be thought to lack some of these features – for example, they do not involve physiological or expressive changes; or more precisely, they involve them only insofar as they involve emotional reactions. In response, Goldie might protest that emotional reactions do not include everything that is distinctive of emotion. Extended emotional attitudes and episodes can involve long term changes to the subject’s motivations and priorities; and they can involve various extended processes of thought which do not have time to develop during a brief emotional reaction. Moreover, like the essentialist strategy, this seems to be a weak principle on which to decide Q3: the birdpox rash may be distinctive of birdpox; but this is no reason to identify birdpox with the rash. 23 A better option for Ekman might be to employ the explanatory strategy. He might argue, in opposition to Prinz, that it is the shape of an emotional reaction that determines how emotional episodes or attitudes develop.13 Even if this were true, however, Ekman would face just the same problem as Prinz: if the coherence claim is correct, the structure of emotional episodes and attitudes can also be explained in functional terms. We are trapped in the same impasse as before. 3.2 Pluralism There is a further possibility to consider. A complex process theorist might decide to adopt a form of pluralism, insisting that emotional reactions, episodes and attitudes can all be regarded as constituting instances of emotion. Indeed, some form of pluralism might be thought to offer the complex process theorist the best chance of explaining what people ordinarily mean when they use terms such as ‘anger’ or ‘fear’ to refer to particular psychological events. For example, consider the following sentence: ‘Alice’s anger was short-lived’ Taken out of context, this sentence might be true, regardless of whether the emotional response was a brief reaction, a full-blown episode or an enduring attitude. In some contexts, it might be read as implying that Alice’s anger dissipated almost immediately; but in other contexts, it might mean that her anger dissipated after a few weeks. Earlier, however, I suggested that in investigating Q3, we are seeking an answer that will be helpful in developing a theory of emotion. It is far from clear that pluralism will deliver such an answer. Once again, the difficulty stems from the diversity claim. I have suggested that emotional reactions, episodes and attitudes are interestingly different kinds of response, and that in discussing certain questions about emotion, it often matters which kind of response we have in mind. Gathering emotional reactions, episodes and attitudes together under a single label runs the risk of blurring the 24 distinctions between them, and so overlooking the richness and variety of emotional phenomena. We might begin to think of emotions exclusively as transient irruptions into the normal course of life, which have little to do with the subject’s beliefs or values; conversely, we might begin to think of them exclusively as extended psychological conditions, apt to exercise a profound influence on the subject’s values and outlook. Complex process theorists, such as Goldie and Ekman, have resisted adopting a pluralistic view, partly because (I take it) they recognise the need to distinguish between different kinds of emotional response. I agree that the distinction is well worth drawing. Where I disagree is over the suggestion that one of these phenomena has more claim than another to be viewed as an instance of emotion. 4. Getting emotional I have been arguing that a principled answer to Q3 may be very hard to find: as theorists of emotion, we might do better to abandon the search. We might wonder whether we should be concerned by this result. I would like to end by suggesting, very briefly, that we should not, for there is an alternative approach to the problem. Rather than arguing over what constitutes an instance of emotion, we might instead adopt a richer vocabulary: one that distinguishes between emotional reactions, emotional episodes, emotional evaluations, emotional feelings, emotional motivations, and so on. There are plenty of questions to ask about the nature of these phenomena, without having to worry about which of them should be honoured with the title ‘the emotion.’ My positive proposal, then, is that we might decide to do without emotions (in the sense assumed by Q3) and become emotional instead. There may be other terminological options. Louis Charland (2010) has recently argued that we should reclaim a distinction between emotions (conceived as transient responses) and passions (conceived as enduring attitudes). This suggestion has the 25 advantage of once established usage. On the other hand, my adjectival proposal avoids any risk of herding types of affective response into separate camps, creating a misleading distinction between enduring passions, such as love or hate, and transient emotions, such as fear and shame. For while there are some types of response that are paradigmatically experienced as enduring attitudes, most can be manifested both as transient and as enduring responses. At any rate, reserving the term ‘passion’ for what I have called an emotional attitude would still leave emotional reactions, episodes, and their various components vying for the title ‘emotion’. My adjectival solution, in contrast, has the advantage of economy. It might be asked, though, what emotional phenomena have in common: what is it that justifies applying the label ‘emotional’ to such a diverse collection of phenomena? Given my response to Q3, how could we answer Q1? My aim in this paper has been to respond to Q3, not to Q1 or Q2. Nevertheless, it is legitimate to wonder whether my response to Q3 leaves scope for a plausible answer to these other questions. The issue is complicated by the fact that, however we decide to tackle Q3, there are separate reasons for doubting whether Q1 has a straightforward answer. As Amélie Rorty (1982: 1) puts it, the term ‘emotion’ may not name a natural class: it may turn out that different types of emotion – anger, disgust, anxiety, love, hope, and so on – are linked only by resemblance. This is an issue that arises quite independently of our answer to Q3. To avoid this difficulty, I shall start, not with Q1, but with Q2. Rather than asking what it is that emotional phenomena have in common, I shall ask what angry phenomena have in common. My aim is not to provide a developed or definitive answer to this question, but to establish that there is space for a plausible answer. Angry reactions, episodes and attitudes, I have argued, share a certain structure and function. Moreover, these different kinds of response are intimately connected: an angry episode characteristically develops from an angry reaction; an angry attitude 26 involves specific episodes of anger. These similarities and connections, I would suggest, are strong enough to justify placing these phenomena in the same psychological category. In contrast, angry evaluations, thoughts and feeling do not themselves have these structural and functional properties. Rather, it makes sense to classify them as angry phenomena because they are components of complex processes that do. As we have seen, many of these phenomena characteristically occur as part of an angry response. This is true of angry feelings, for example. It is true because the phenomenal flavour of such feelings is characteristically explained by the distinctive mix of psychological and physiological changes that comprise an angry response. For this reason, it might make sense to describe such a feeling as an angry feeling even if – as the result of some malfunction, say – it were to occur in isolation. For it is just the kind of feeling that we characteristically experience when we are angry.14 In contrast, the thought ‘He did that on purpose!’ might occur either within an angry episode or as a calm assessment of the situation. In the latter case, there is no reason to describe it as an angry thought. But in the former, it does make sense to characterise it as an angry thought: to do this is to locate it within a particular kind of psychological process, explaining why it occurred on this occasion and helping us to predict its likely effects. On this account, then, there is no single reason for classifying something as an angry phenomenon. Rather there is a clutch of closely related reasons, all depending on the idea that angry responses share a certain structure and function. It should be easier now to see how we might respond to Q1. The idea will be that anger, fear, love, and so on can be classified as types of emotion, not because they share some common feature, but because their structural properties are recognisably similar; and because, in many cases, they have similar kinds of function. Insofar as this is true of anger, it will be right 27 to describe angry episodes, feeling and thoughts as emotional episodes, feelings and thoughts. My response to Q3, then, does not preclude a response to Q1 and Q2. I have argued that we might do well to abandon our search for a principled answer to Q3, and instead adopt my adjectival solution. As I have explained, though, my argument is limited in two ways. First, I have considered a finite number of strategies for answering Q3: it is always possible that I have overlooked some reasonable strategy that would yield a decisive and theoretically interesting answer to Q3. Secondly, my argument rests on a particular conception of emotional responses – as diverse, complex and coherent. I have acknowledged that the coherence claim, in particular, is in need of further defence. Until that is supplied, there remains the possibility that the explanatory strategy might yield a reasonable, decisive and interesting answer to Q3. Nevertheless, I believe that the picture of emotion that I have presented here is a plausible and powerful one. If it is correct, Q3 may well turn out to be a less significant question than we might have thought.15 Department of Philosophy The Open University, UK Notes 1 Ekman, 1992; Aaron Ben-Ze’ev (2010: 46) has recently defended the view that instances of emotion are characteristically brief. 2 For more detailed characterisations of enduring emotional states, see Goldie, 2000; Charland, 2010. 3 It would be possible to hold that emotional responses can be given a functional analysis, while rejecting the view that claims about function should be understood in historical terms (Delancey, 2002). However, it is essential to my argument that the 28 function of an emotional response explains why it has a certain structure, and for this, I would suggest, the claim about function needs to be understood as a historical claim. 4 For some different views on this, see Averill, 1980; Tooby and Cosmides, 1990: 407- 408; Ekman, 1992: 171 Griffiths, 1997. 5 My argument does require that these are the exception, rather than the rule. However, I believe that many apparently difficult cases, such as grief or love, can plausibly be understood in functional terms. (For an account of grief see [Author, 2010]). 6 It might be suggested that Roberts’ seventh claim rules out a complex process theory. However, it is not clear that a complex process theorist needs to reject this claim: complex emotional responses, I have suggested, are experienced as a unified and coherent package of response, not as a collection of disparate symptoms. 7 This is not to deny that there are further moves that can be made here. Indeed, Roberts himself makes further moves. In doing so, however, he employs a different strategy, which I shall discuss in section 2.4. 8 The exception here may be emotional behaviour: it does seem to be possible to let one’s anger blossom, while keeping one’s behaviour in check. And indeed, it does seem more natural to insist that S1 is a causal explanation. 9 A proponent of this account will need to explain why S6 presents Alice’s evaluation as explaining her anger rather than the other way around. 10 This is a constitutive rather than a functional explanation. The function of an emotion becomes relevant when we ask a different question: why emotional responses of a given type characteristically have the structure that they do. 11 In section 1.2, I raised a concern about Prinz’s explanation of how the value of the valence marker gives rise to emotional motivations and behaviour. 29 12 Similarly, the fact that human beings have hearts can be explained both causally (by genetic and developmental facts) and functionally. 13 Which view is correct is an empirical issue: the issue turns on whether an emotional reaction should be regarded as pre-programmed response which is simply triggered by an emotional appraisal or evaluation or whether the nature of the evaluation plays a significant role in determining how the reaction develops. 14 Similar points might be made about angry evaluations, on the assumption that these states too characteristically occur as part of an angry response and have a distinctive function to play. 15 I have presented earlier versions of this paper at a seminar at the Open University and at a meeting of the Mind and Reason group at the University of York. I am very grateful to the participants at those events for their valuable comments. I would also like to thank an anonymous referee for this journal for some very helpful suggestions. References Averill, J. (1980). ‘A Constructivist View of Emotion,’ in R. Plutchik and H. Kellerman (ed.s) Theories of Emotion. New York: Academic Press, pp. 305-339. Ben-Ze’ev, A. (2010). ‘The Thing Called Emotion’ in P. Goldie (ed.) The Oxford Handbook of Philosophy of Emotion. Oxford: Oxford University Press, pp. 41- 62. Charland, L. (2010) ‘Reinstating the Passions: Arguments from the History of Psychopathology’ in P. Goldie (ed.) The Oxford Handbook of Philosophy of Emotion. Oxford: Oxford University Press, pp. 237-259. Delancey, C. (2002). Passionate Engines. Oxford: Oxford University Press. de Sousa, R. (1987). The Rationality of Emotion. Cambridge MA: MIT Press. 30 Ekman, P. (1992). ‘An Argument for Basic Emotions,’ Cognition and Emotion 6, pp. 169-200. Goldie, P. (2000). The Emotions: A Philosophical Exploration. Oxford: Oxford University Press. Griffiths, P. (1997). What Emotions Really Are. Chicago: University of Chicago Press. Helm, B. (2001). Emotional Reason: Deliberation, Motivation and the Nature of Value. Cambridge: Cambridge University Press. James, W. (1890). The Principles of Psychology, Vol. II. New York: Dover. Kenny, A. (1963). Action, Emotion and Will. London: Routledge & Kegan Paul. Lyons, W. (1980). Emotion. Cambridge: Cambridge University Press. Marks, J. (1982). ‘A Theory of Emotion’, Philosophical Studies 42, pp. 227-42. Nussbaum, M. (2001). Upheavals of Thought: The Intelligence of Emotions. Cambridge, Cambridge University Press. Price, C. (2010). ‘The Rationality of Grief,’ Inquiry 53, pp. 20-40. Prinz, J. (2004). Gut Reactions: A Perceptual Theory of Emotion. Oxford: Oxford University Press. Roberts, R.C. (1988). ‘What an Emotion Is: A Sketch,’ The Philosophical Review 97, pp.183-209. Robinson, J. (2005). Deeper Than Reason: Emotion and its Role in Literature, Music and Art. Oxford: Oxford University Press. Rorty, A. (1982). ‘Introduction,’ in A. Rorty (ed.) Explaining Emotions. Berkeley: University of California Press, pp.1-8. Schaffer, J. (1983). ‘An Assessment of Emotion,’ American Philosophical Quarterly 20, pp.161-173 31 Solomon, R. (1993). The Passions: Emotions and the Meaning of Life. 2nd ed. Indianapolis: Hackett. Stocker, M. (1987). ‘Emotional Thoughts,’ American Philosophical Quarterly 24, pp. 59-69. Tooby, J. and Cosmides, L. (1990). ‘The Past Explains the Present: Emotional Adaptations and the Structure of Ancestral Environments,’ Ethology and Sociobiology 11, pp. 375-424. 32