Getting Bodily Feelings Into Emotional Experience in the Right Way

The Objection From Dispositions

Several prominent philosophers of emotion have argued that there is no necessary relation between emotions and feelings, bodily or otherwise. Consider cases such as loving someone or being ashamed of one’s social background. Can’t we be in such states without feeling anything (Bedford, 1957; Pitcher, 1965; Ryle, 1949; Smith, in press)? It seems that we can. After all, one can be truly said to love someone or be ashamed of one’s background even when one is fast asleep or when one’s mind is busy solving a mathematical problem. Shame and love can last for years and it needs not feel like anything to be in those states.

This particular version of the objection is now widely regarded as inconclusive, however (e.g., Alston, 1969; Myers, 1969; Perkins, 1966). The rejoinder consists in drawing attention to the distinction between dispositional states and episodes. James (1890/1950) no doubt regarded emotions as mental episodes and thus meant specificity to concern emotional episodes rather than emotional dispositions. Emotional episodes are, typically at least, manifestations of corresponding dispositions—for example, episodes of shame at having to tell where one comes from manifest one’s dispositional shame about one’s social background. Emotional dispositions exist, and they are indeed not marked by feelings. Still, it is difficult to deny that there are also emotional episodes, and pointing out that emotional dispositions are not associated with feelings is no reason for thinking that emotional episodes are not so associated.

The claim that there is no significant relation between emotions and feelings may have a different source, however. One can grant that emotional episodes feel like something, yet deny that different emotions are associated with different feelings (e.g., Kenny, 1963; Lyons, 1980; Robinson, 2005). Consider for instance love and ask yourself what a “feeling of love” consist of. According to Kenny and Lyons, love does not have a privileged link to any particular kind of feeling. Rather, feelings of almost any kind may manifest love in the relevant circumstances: a pang of jealousy, a feeling of amusement at the social blunder of the beloved person, and a feeling of fear for her safety may all do so. Conversely, it is argued, one and the same feeling can manifest very different emotional states. The episode of amusement may manifest one’s love for the person having committed the blunder, but it may also manifest contempt for the norm of etiquette flouted by her. This is taken to imply that the expression “feeling of love” does not refer to a specific kind of feeling, but serves instead to refer to the feelings, which may be of many sorts, that occur because of love. This conclusion is then generalized to other emotions (e.g., Kenny, 1963, pp. 98–99; Lyons, 1980, 130ff.; Robinson, 2005; Ryle, 1949): there are no emotion-specific feelings of shame, admiration, fear, hope, or anger, only less specific feelings whose occurrence is sometimes explained by these emotions. For example, the feelings of shame may comprise the feeling of fear of having to reveal your background during an interview, the feeling of hope that this will not be the case, and the feeling of anger at someone keen on reminding you of where you come from.

This argument is most likely correct in emphasizing that some emotions, such as love, are not associated with specific types of feelings. Its conclusion must nevertheless be resisted, because it rests again on the failure to draw the distinction between emotional dispositions and episodes. ² Love is what philosophers sometimes call a “multi-track disposition” (Ryle, 1949): an emotional disposition which can manifest itself in a variety of emotions depending on the circumstances. ³ Moreover, many or even most of the words used to refer to emotional episodes (being ashamed, angry, afraid, etc.) can also be used to ascribe emotional dispositions. In these cases, there is an explanatory relation between the relevant emotional disposition and the feelings associated with the corresponding emotional episodes, and it may well be that in some cases such as love, these feelings are not specific to the emotional disposition. What is at issue, however, is whether this conclusion carries over to all emotions. Even if we accept that episodes of love are not marked by distinctive feelings of love, and that other emotional dispositions such as shame can manifest themselves in a variety of feelings, do we need to conclude that there are therefore no feelings specific to shame, amusement, fear, or anger? No. The fact that we sometimes use “being ashamed,” “angry,” or “afraid” to refer to dispositions does nothing to show that, when we use them to refer to episodes of shame, anger, or fear, the feelings associated with these episodes are not specific to them.

In sum, the objection we have considered does not support the conclusion that specificity is false; it only supports the conclusion that specificity is restricted to emotional episodes. While emotional dispositions should be excluded from the scope of the specificity claim, nothing in the foregoing shows that specificity does not hold true for emotional episodes.

Before we move on, we would like to point out that the insight underlying the first objection (that emotion words are often used to ascribe emotional dispositions) has led some philosophers to reach a different conclusion, not regarding the specificity of emotional feelings, but the nature of emotions. It has been argued that the attribution of emotional dispositions features so prominently in our practice of attributing emotions that feelings, however specific they are, are not sufficient for defining emotions. This consideration has nourished an important minority view, according to which emotions are multitrack dispositions not only to feel, but also to think, imagine, behave, etcetera in distinctive ways (Goldie, 2000, 2011; Wollheim, 1999). Ryle (1949) may in fact have held this view himself. Whether this theory of emotions is true or not, the important point at this stage is that it is compatible with the idea that emotional dispositions manifest themselves in feelings that are type-specific. ⁴ So, let us turn to the second objection against specificity.

The Objection From Arousal Misattribution

The results of Schachter and Singer’s (1962) famous experiment on the misattribution of arousal have been interpreted by some authors as showing that there is no specificity to the bodily feelings associated with emotions (e.g., Lyons, 1980, Chapter 8). Let us briefly remind ourselves of this experiment before examining why it may support an argument against specificity.

Conceived in the most general terms, Schachter and Singer’s (1962) data seem to show that subjects who share the same bodily feelings of general arousal can, by being placed in different social settings, come to interpret these feelings as (part of) distinct emotions. In the experiment, all participants except for controls were injected with the same amount of epinephrine, this raising their state of arousal to more or less the same degree. The different settings consisted of the presence of stooges who created either a positive or a negative atmosphere by enacting different social scripts in the presence of the participants. Results appeared to show that, when participants were uninformed or misinformed about the effect of the adrenaline injection, they interpreted their bodily feelings in different ways depending on the setting they were in. In a positive atmosphere, they interpreted their bodily arousal as (part of) a positive emotion, whereas in a negative atmosphere, they interpreted it as (part of) a negative emotion.

Schachter and Singer’s (1962) study has been the target of many criticisms, and few now believe that their results can be taken at face value. Insofar as these results do not withstand scrutiny, they do not threaten specificity. ⁵ Still, we believe that it is of considerable interest to assess whether their predictions regarding arousal misattribution would, if confirmed, threaten this Jamesian claim. The reason is that this issue raises a more general, important question: the question of the relation between specificity and self-attributions of emotion.

At first sight, it appears that there is no strong or necessary connection between the two: Specificity is not a claim about how people self-attribute emotions on the basis of what they feel, but a claim about the specificity of their emotional feelings. Why, then, did some authors (e.g., Lyons, 1980), including most likely Schachter and Singer (1962) themselves, think that their findings indirectly show that there are no type-specific emotional feelings? The underlying train of thought may have been the following. Participants to the experiment experienced the same unspecific bodily feelings resulting from the adrenaline injection, yet they self-attributed different emotions in different settings. Furthermore, it is reasonable to suppose that the central determinant of these participants’ self-attributions was their understanding of the cause of their bodily feelings. Now, if this is true in the context of the Schachter–Singer experiment, it is probably also true in ordinary circumstances. Therefore, emotions are constituted by unspecific bodily feelings and an understanding of their likely cause. ⁶

The weak spot of this argument is its assumption that the self-attribution of emotions requires reference to the cause of bodily feelings. The findings of the Schachter–Singer study show at best that such a causal attribution is required when bodily feelings are unspecific, a situation that generates what Schachter and Singer call an “evaluative need” (1962, p. 382). We may agree that a subject who has unspecific bodily feelings can be brought to interpret her arousal in terms of a specific emotion if the circumstances invite such an interpretation. Still, does the fact that subjects sometimes (perhaps even often) use situational cues in addition to their feelings to understand their own emotions refute specificity? No. A more sensible conclusion, we believe, is that self-attributions of emotions are not always and exclusively based on introspection of how the body feels.

Against this, the proponents of the “objection from arousal misattribution” might reply that participants’ self-attribution of emotions on the basis of unspecific bodily feelings is in fact highly diagnostic for how we normally self-attribute emotions. For if emotions were associated with a rich and specific phenomenology, this would certainly shape our practice of self-attribution—we would, typically at least, self-attribute emotions on this basis. If this is how self-attribution normally works, then it would have been salient to the participants of the Schachter–Singer experiment that the normally present, emotion-specific phenomenology was absent. This, in turn, would have blocked, or at least hindered, the self-attribution of a specific emotion. Given that the participants of the experiment did self-attribute specific emotions, it can be concluded that the self-attribution of emotions is not driven by emotion-specific feelings, and hence that no such feelings exist.

Although this reasoning has some plausibility, it rests on two controversial assumptions regarding the roles specific bodily feelings would have if they exist. The first assumption is that such feelings would (alone) determine most self-attributions of emotions. The second is that one could not easily be made to self-attribute emotions in their absence. We believe that both of these assumptions are controversial, because they are difficult to reconcile with a fundamental aspect of emotional experience. As we have emphasized earlier, when one undergoes an emotion, it seems as if one is affectively engaged with something significant in the environment (Goldie, 2000; Tappolet, 2011). Now, it would be strange if what is at the centre of attention during these outward-looking emotional experiences had no impact on self-attribution. Thus, against the first assumption, we believe that it is in fact quite likely that the subject’s evaluation of her environment will have an impact on which emotions she ends up self-attributing, even if she experiences emotion-specific bodily feelings. This in turn suggests, against the second assumption, that the way the person evaluates the situation can sometimes, or even often, bypass bodily feelings and determine the specific emotion she will self-attribute—for example, if she thinks the situation is offensive, she may self-attribute anger even if she feels just bad, or even if she feels nothing at all. These conclusions hold independently of whether or not emotion-specific feelings exist.

Depite the interesting considerations raised by Schachter and Singer’s (1962) work, our conclusion therefore mirrors that reached in the previous section. There is no easy and direct route from emotion attributions to the nature of emotional feelings, and certainly none that requires us to reject specificity.

The Objection From Fine-Grainedness

The third objection against specificity is based on a score of largely uncontroversial facts about ordinary language. One such fact is the cultural variability of emotion categories, as evidenced by surveys of different languages (e.g., Ogarkova, 2013; Russell, 1991; see also Scarantino, 2014a). Another acknowledged fact about the language of emotion is the fine-grainedness of the emotion categories present in many languages. For example, we speak of anger, frustration, irritation, annoyance, rage, exasperation, etcetera. Such rich semantic fields exist for most families of emotions. The objection from fine-grainedness is that it is difficult to believe that specific bodily feelings correspond to all these fine-grained emotion categories. If these different categories are not associated with specific patterns of bodily feelings, it is insisted, specificity is refuted.

We concede that the way in which the Jamesian partitions the emotional space with the help of specificity is unlikely to mirror the fine-grainedness of folk categories of emotion. How damaging is this concession, however? An obvious move for advocates of specificity is to claim that only a privileged subset of emotions is associated with specific bodily feelings. Furthermore, the advocates of specificity can claim that these priviledged emotions need not closely mirror the emotion categories demarcated in ordinary language.

As is well known, the first assumption is the central tenet of a prominent family of emotion theories, the basic emotion theories. ⁷ Basic emotions are typically conceptualized as evolutionary modules activated by circumscribed sets of stimuli signaling fundamental life tasks, whose function is to coordinate emotion-specific responses in a way best suited to deal with these tasks. Central to basic emotion theories is the hypothesis that emotions are characterized by distinct patterns of changes in facial and skeletal muscles, and in the autonomic and endocrine systems (e.g., Ekman, 1980, 1999; Izard, 1977, 1992; Scarantino, 2014a). Whether these privileged sets of emotion-specific patterns of bodily changes correspond to the emotions denoted by familiar emotion words (fear, anger, sadness, etc.), or whether these words refer to families of distinct patterns of bodily changes that share some similarities, is to be decided empirically (see Prinz, 2004, Chapter 6 and Scarantino, 2014a, for discussions).

The prospects of specificity then depend on the plausibility of the claim that the emotional domain is structured around a number of basic emotions. More specifically, within the Jamesian framework, its prospects depend on the assumption that at least the basic emotions are associated with differentiated bodily responses. How plausible is this assumption? As a matter of fact, it has in recent years been the target of numerous criticisms (e.g., Barrett, 2006a, 2006b; Lindquist, Wager, Kober, Bliss-Moreau, & Barrett, 2012; Russell, 2003). The critics insist that there are no specific patterns of bodily changes associated with the basic emotions, and even that emotions can occur in the absence of bodily feedback (see Reisenzein & Stephan, 2014, for a review). The evidence for the existence of emotion-specific patterns of physiological reactions, facial expressions, and even feelings (see e.g., Kreibig, 2010; Laird & Lacasse, 2014; Levenson, 1992; Stephens, Christie, & Friedman, 2010) is claimed by the critics to be fragmented and controversial. To this, friends of basic emotions reply that the fragmentation of the evidence could be due to method problems (see Mauss, Levenson, McCarter, Wilhelm, & Gross, 2005 and Reisenzein, 2000, for discussions). This is not the place to adjudicate between these positions. However, because the empirical evidence is apparently not conclusive either way, it would be an overreaction to abandon specificity.

There is, however, a second, theoretical objection against basic emotion theory. This objection is that the sort of phenomena basic emotions are claimed to be cannot be made to fit the functional role of emotions (e.g., Griffiths, 1997). In a nutshell, the charge is that basic emotions are much too rigid to fit that role. Consider what happens “upstream” of basic emotions. They are said to be automatically activated by a restricted class of stimuli, those that signal the presence of a fundamental life task. Yet, our practice of ascribing emotions indicates that we assume that emotions of the same type can be triggered by very diverse stimuli. These stimuli range from the purely sensory to the cognitively elaborated, and may have quite indirect connections to fundamental life tasks. If basic emotion theory can only account for emotions elicited by simple stimuli, then its explanatory capacity is severely diminished. Griffiths (1997), for one, bites the bullet and argues that radically different explanations must be given for those phenomena that we call emotions in everyday life but that do not satisfy the requirements of basic emotions (e.g., fear elicited by the thought of having lost one’s fortune in stocks).

Consider next what happens “downstream.” It is usually held that the output of basic emotion modules unfolds in a cascade-like manner, that is, automatically and rigidly (e.g., Ekman, 1980, 1999). Here, the worry is that, as judged by our practice of ascribing emotions in everyday life, people having the same emotion can respond in very different ways, and these different ways are typically modulated by the circumstances. For example, an angry person could physically attack her opponent, or she could write an insulting letter. Again, as long as basic emotion theory is committed to positing rigid emotional responses, its explanatory power is doubtful (Scarantino, 2014a).

Are we here facing a serious objection against basic emotion theory and, as a result, against specificity? We do not think so. While upstream and downstream rigidity have been emphasized by some advocates and critics of basic emotion theory, the basic emotions research program can make room for flexibility at both the input and the output side of the emotion modules. In fact, the assumption that such flexibility exists is now being made by many proponents of the basic emotions view (e.g., Ekman & Cordaro, 2011; Izard, 2011; Levenson, 2011; Scarantino, 2014a).

Specifically, to address the upstream worry, one needs to assume that basic emotions can also be activated by cognitively more or less complex psychological states (e.g., Arnold, 1960; Deonna & Teroni, 2012; Lazarus, 1991; Prinz, 2004, Chapter 6). Many of the distinctions between emotions that we make in ordinary language will thus be the consequence of fine nuances in the type of evaluations involved (e.g., while frustration is directed at perceived obstruction, anger is directed at perceived offense, and indignation at perceived injustice) rather than fine nuances in bodily feelings. This is to say, then, that there may be a subset of emotion elicitors that is, for evolutionary reasons, privileged, but the set of potential elicitors can subsequently be enlarged by association, learning, deliberation, and more generally, concept acquisition (Prinz, 2004; Roseman, 2011).

To address the downstream worry, one needs to allow that emotional responses can be modulated by circumstances. In fact, if there is much variety in the potential elicitors of any given emotion, then this variety should be expected to be reflected in the different ways the emotional response may unfold. Emotional responses are not rigid, but something much more open-ended and flexible. At the same time, however, one must be careful not to conceptualize emotional responses as too open-ended, on pain of playing into the hands of the critics of specificity. The most promising option is, we believe, to appeal to states of action readiness. States of action readiness feature already in McDougall’s (1923), Bull’s (1951/1968), and Arnold’s (1960) emotion theories, they have been systematically explored by Frijda (1986, 2007), and they have more recently also been put to use by philosophers (Deonna & Teroni, 2012; Scarantino, 2014b).

States of emotional action readiness are defined by what Frijda (1986) calls a relational goal taking control precedence. There is a preparedness to act in a particular direction which harnesses the subject’s resources at the expense of her other goals and/or activities (Scarantino, 2014b, p. 170). This preparedness to act is sustained by changes in a variety of organismic systems including facial and skeletal muscles, the autonomic nervous system (e.g., heart rate, perspiration, respiration, digestion), and the endocrine system (e.g., adrenaline). These changes do not only happen, but are, Jamesians will insist, typically felt and so contribute to shaping the distinctiveness of emotional experiences. In fear, we feel the way our body is poised to act so as to contribute to the neutralization of what provokes the fear. In anger, we feel the way our body is prepared for active hostility to whatever causes the anger. In shame, we feel the way our body is poised to hide from the others’ gaze, which typically causes the shame. In an episode of loving affection, we feel the way our body is prepared to move towards cuddling the object of one’s affection. In sadness, the body is felt as though it were prevented from entering into interaction with a certain object, etcetera. ⁸

The descriptions of bodily reactions suggested by the action readiness theory—if they capture the way emotions manifest themselves in consciousness—appear to satisfy a double constraint. ⁹ First, they are still diverse enough to demarcate at least a significant subset of emotions for which specificity may hold. Second, they are open-ended enough to avert rigidity-related worries. All in all, then, the concession that specificity holds only for a subset of folk emotion categories might not be too much of an obstacle. Appeal to felt states of action readiness that can be activated by a wide range of elicitors and that are modulated by the circumstances seems to be a viable account of emotional experience that is fundamentally Jamesian in spirit.

The Objection From Action Explanation

One reason for thinking that bodily feelings, even if they are emotion-specific, are not constitutive of emotions is the following: bodily feelings must be kept out of the emotions to preserve our intuition about the explanatory relation that exists between emotions and bodily feelings. In line with this intuition, bodily feelings are caused by the emotions, and for that reason are not part of them (e.g., Gordon, 1987; see also Deigh, 2014). There is indeed much to be said in favour of what we can call the pre-Jamesian sequence, according to which a stimulus causes an emotion, which in turn causes bodily changes that may be felt. If this intuition is correct, then emotional phenomenology cannot be understood in terms of bodily feelings and our efforts to defend specificity have failed to reach their ultimate, Jamesian goal: bodily feelings are not constitutive of the emotions, but consequences of them.

To handle this objection, we need to become clearer about the exact nature of the pre-Jamesian claim. It is not simply that some bodily feelings must be explained by emotions. On this, all can agree. To give one example, the feeling of one’s fist going in the direction of the offender’s face is plausibly a feeling that is caused by anger rather than part of it. The disagreement concerns the question of whether the whole felt bodily dimension of emotions must be counted amongst their effects.

If it is so counted, then we face a dilemma brought out by James’s subtraction argument. We can either accept the consequence of James’s argument, that is, that nothing phenomenological is left once bodily feelings are factored out of the emotions. In this case, we need to conclude that emotions have no phenomenology. Alternatively, we can reject the consequence of James’s argument and maintain that something phenomenological is left after bodily feelings have been factored out. This is the horn of the dilemma that Goldie (2000) chooses when he insists that emotions involve nonbodily feelings. In this case, we can conclude that emotions do have a phenomenology, but we are then required to explicate the nature of the nonbodily feelings that constitute emotions. Goldie (2000) answers this challenge by referring to the nonbodily feelings at stake as “feelings towards,” whose essential function is to make significant aspects of the environment manifest to us.

Presented in this way, neither horn of the dilemma is attractive. Going for the first horn and accepting that emotions have no phenomenology must be the last resort. As argued in what follows however, accounting for the explanatory role of emotions does not force us to take this course. As for the second horn of the dilemma, accepting the existence of Goldie’s “feelings towards” requires positing an emotional phenomenology whose nature remains mysterious and whose existence appears to be confirmed neither by introspection nor by more controlled methods of investigation. These observations provide an incentive for reconsidering the causal sequence assumed in the pre-Jamesian claim. Does the preservation of a satisfactory explanatory role for the emotions really require to either deny that emotions have a phenomenology or to appeal to mysterious “feelings towards”? There are at least two strategies for answering this question negatively.

The first strategy consists in the partial acceptance of the subtraction argument. Stripping emotions of their entire bodily dimension is, one agrees, to be left with something that cannot be recognized as emotion any more. So one tries to strike a middle ground: there is a felt bodily dimension to the emotions, but there is also the need to explain action by the emotions. This suggests that those aspects of emotional phenomenology that are action-oriented should be factored out of the emotional experience. As Russell puts it, emotions only include “a general mobilization for vigorous action” (2003, p. 150). Such a “partitioning” of emotional phenomenology is salient in contemporary accounts of emotional experience that, influenced by Schachter and Singer’s (1962) work, portray emotional feelings as mixtures of valence (positive or negative affect) and arousal (high or low bodily activation)—often called ”core affect” (Barrett, 2006a, 2006b; Russell, 2003). We are left with an impoverished phenomenology that needs cognitive supplementation to constitute specific emotions.

This way of circumventing the dilemma comes at a price, however. It satisfies constitution, but we believe it does so in a way that gives up the idea that emotional experiences are evaluative attitudes. That is, the core affect theory seems to miss the fact that, in emotions, the phenomenology is that of seeming to be related to a situation that has a distinctive significance. ¹⁰ Furthermore, we believe that one motivation for core affect theory—that there is no one-to-one correspondence between emotions and specific behavioural outputs, as claimed by basic emotion theory—can be acknowledged without having to adopt that theory (see the The Objection From Evaluation section).

The second strategy for accommodating the sort of explanations manifest in the pre-Jamesian sequence is to appeal to states of felt action readiness. We have already seen that states of felt action readiness allow us to preserve specificity and to accommodate emotional flexibility. If so, the only remaining issue is whether the assumption that emotions are states of felt action readiness is compatible with the explanatory role we want to preserve for the emotions. In particular, is it acceptable to give up the possibility to explain states of felt action readiness by reference to emotions? We think it is, for two reasons. First, doing so does not conflict with the pre-Jamesian intuition that emotions are psychological intermediaries between stimuli and action. Appealing to action readiness does not prevent us from continuing to say, for example, that one may flee out of fear, cajole out of affection, strike out of anger, etcetera. Second, we have surveyed the available alternatives and hope to have shown that higher prices attach to them.

The Objection From Evaluation

The assumption that emotional experiences are states of felt action readiness may be compatible with preserving the explanatory role of the emotions. Yet, this assumption may make it difficult to capture the evaluative dimension of emotions. Why?

States of felt action readiness are, we argued, specific types of bodily feelings. And, as the expression indicates, bodily feelings are feelings of the body. We have repeatedly emphasized, however, that the phenomenology of emotion is that of apprehending the significance of something. In fear, a situation is apprehended as threatening, in anger, it is apprehended as offensive, and so on. These evaluations, furthermore, lend themselves to a variety of normative assessments: emotions can be reasonable or unreasonable, more or less fitting to their objects, and so on (e.g., D’Arms & Jacobson, 2000; Kenny, 1963). For example, philosophers have famously argued that it is unreasonable to fear death, and that death is not a fitting object of fear. If the phenomenology of emotions essentially involves apprehensions of significance, however, then Jamesian accounts (of any type) seem to have a problem. Bodily changes are obviously not what is presented to the subject as being significant for her in emotions, and bodily feelings do not lend themselves to the described normative assessments. Therefore, it seems as if emotions cannot be constituted by bodily feelings after all.

A traditional way of escaping this conclusion consists in conceiving of bodily feelings as one component of emotional experience amongst others. Bodily feelings, it is argued, account for the phenomenological dimension of emotions, while a cognition accounts for their evaluative dimension. All theories that conceive of emotions as combinations of components favour such an approach (e.g., Gordon, 1987; Lyons, 1980; Russell, 2003; Scherer, 2009). This includes some interpretations of Schachter’s theory (see Reisenzein, 2017). These theories, however, face the objection that they are “add-on” theories (Goldie, 2000), that is, they simply put distinct cognitive and feeling components alongside each other. Add-on theories are seen as unappealing primarily because they conflict with the intuition that the significance revealed by emotions is intimately bound to the way it feels to experience them. Consider what happens in visual perception: the way a surface’s redness is revealed to you cannot be divorced from the way it feels to have the visual experience. Similarly, it is claimed, the disgusting aspect of an object revealed to you in disgust is fundamentally tied to the way it feels to experience disgust. Intentionality and phenomenology being so intertwined, it is argued that it is hopeless to try to distribute the cognitive and phenomenological dimensions of emotions over two separate components. As Goldie (2009) emphasizes, feelings “are bound up with the way we take in the world in emotional experience” (see also Leighton, 1984; Pugmire, 1998, p. 58; Ratcliffe, 2008, Chapter 1). ¹¹

If adding evaluative cognitions to bodily feelings fails to do justice to the intertwinement of phenomenology and intentionality characteristic of emotions, are we forced to abandon constitution? The only alternative seems to be to assume that bodily feelings themselves account for the evaluative dimension of the emotions. But how can an outward-looking evaluation be accounted for in terms of bodily feelings? After all, according to a traditional way of thinking about it, the bodily phenomenology in emotion manifests itself when we introspectively “look inward.” And true enough, introspection results in the mere awareness of changes in a variety of bodily parts. If this is how the body feels in emotion, then constitution should be abandoned.

However, does introspection faithfully reveal the bodily phenomenology of emotions? The answer seems to be that, on the contrary, inward-looking attention actually disrupts the intentional structure of bodily phenomenology. Here is how Claparède (1928), an early supporter of James, expressed dissatisfaction with the way the latter conceived of bodily phenomenology: “Permit yourself to become absorbed in [the object of] your anger; . . . then you no longer experience distinctly the trembling of your lips, your pallor, or the isolated sensations arising from the different parts of your contracted muscular machinery” (1928, p. 129).

According to Claparède, in emotional experience the subject’s attention is typically directed outward. ¹² When this is the case, he argued, the body feels altogether differently, that is, it is experienced in a holistic rather than atomistic fashion: “The emotion is nothing other than the consciousness of a form, of a ‘Gestalt’ of these multiple organic impressions. . . . This confused and general perception of the whole . . . is the primitive form of . . . emotional perception.” (1928, p. 128).

This is evocative, but what exactly does it mean to say that the body is felt “in a holistic way”? Turning to Frijda (2005) can again help:

Intentional nature and meaning [of emotion experience] depend on the current mode of attention. They are most distinct in synthetic mode and immersion and they can be destroyed with increasing degree of analyticity and detachment. In self-focus, analytic attention reduces felt bodily engagement to just that. Felt impulse to shrink back from a threat is transformed into felt muscle tension, just as the feeling of pointing can be transformed into feeling one’s finger stretched. One comes to feel dizzy, one’s heart racing, instead of feeling anxious or upset. . . . Analytic isolation of information sources robs an experience of its emotional character. (2005, p. 482)

In this passage, Frijda (2005) argues that states of action readiness actually appear in consciousness as ways of being engaged with the environment. If this is correct, then states of action readiness do have the appropriate sort of directionality, that is, they are outward-looking. States of action readiness are attitudes or stances that we adopt in relation to external objects or situations, and this is precisely how they feel.

If we accept this, the final question we need to answer is whether the outward-looking intentionality of states of action readiness also qualifies as evaluative. In other words, can states of felt action readiness explain the evaluative dimension of emotions? A positive answer requires that we clarify the way in which emotions evaluate their objects and the normative assessments to which they are subject in terms of felt bodily stances. Here is our proposal. Fear is a danger-related evaluative experience because it consists in feeling one’s body’s readiness to act so as to diminish an object’s likely impact on one (flight, preemptive attack), and this felt attitude is both fitting and reasonable if the dog is dangerous. Fear is subject to normative assessments of this kind precisely because it is such a distinctive attitude towards an object. Similarly, anger is an offense-related evaluative experience because it consists in feeling one’s body’s readiness to act so as to retaliate one way or another, and this felt attitude is both fitting and reasonable if there has been an offense. Anger is subject to normative assessments of this kind precisely because it is another distinctive felt bodily attitude towards an object. ¹³

If this proposal is along the right track, then felt bodily stances allow the Jamesian to defend a sensible version of both specificity and constitution that also illuminates the evaluative dimension of emotions.