The reactive theory of emotions

Evaluative theories of emotions purport to shed light on the nature of emotions by appealing to values. Three kinds of evaluative theories of emotions dominate the recent literature: the judgment theory equates emotions with value judgments; the perceptual theory equates emotions with perceptions of values, and the attitudinal theory equates emotions with evaluative attitudes. This paper defends a fourth kind of evaluative theory of emotions, mostly neglected so far: the reactive theory . Reactive theories claim that emotions are attitudes which arise in reaction to perceptions of value.


| INTRODUCTION
Evaluative theories of emotions purport to shed light on the nature of emotions by appealing to values.Three kinds of evaluative theories of emotions dominate the recent literature: the judgment theory equates emotions with value judgments; the perceptual theory equates emotions with perceptions of values, and the attitudinal theory equates emotions with evaluative attitudes.This paper defends a fourth kind of evaluative theory of emotions, mostly neglected so far: the reactive theory.Reactive theories claim that emotions are attitudes which arise in reaction to perceptions of value.As a first approximation, here is how each theory accounts for the prototypical case of the fear of a tiger: • Judgment theory: to fear the tiger is to judge it to be dangerous.
• Perceptual theory: to fear the tiger is to experience it as dangerous.
• Attitudinal theory: to fear the tiger is to have a fear-attitude toward the tiger, which is correct if and only if the tiger is dangerous.
• Reactive theory: to fear the tiger is to have a fear-attitude towards the tiger in reaction to its being apparently dangerous.
All these theories are evaluative theories of emotions in the sense that a value-term-here "danger"-figures in the explanans. 1Not all of these theories, however, are reductive: "fear" appears in the explanans of both the attitudinal and the reactive theories.Both theories are in this respect primitivist about emotions: contrary to the judgment theory and to the perceptual theory, the attitudinal and the reactive theories do not attempt to analyse emotions.Rather they specify the category to which they belong (attitudes) and spell out their relation to values (in terms of correctness and/or reaction).
I shall here focus on the three last theories: the perceptual, the attitudinal and the reactive theories.Each has ancestors in Brentano's school.Meinong (1917) developed a version of the perceptual theory, Brentano (1995) endorsed a version of the attitudinal theory and the reactive theory was defended by Reinach (1989), von Hildebrand (1916, 1922), and Scheler (1973).Admittedly, the perceptual and the attitudinal theories are the most influential nowadays.The reactive theory has been largely-albeit not totally-ignored in the analytic tradition.This article argues that the reactive theory is to be preferred to both the perceptual and the attitudinal theories.My argument in a nutshell is that only the reactive theory can account for the fact that reasons for entertaining emotions are evaluative: the reason for which we fear the tiger is that it is dangerous.Both the perceptual and the attitudinal theories, by contrast, end up saying that the reason for which we fear the tiger is that it has sharp teeth-a non-evaluative property.
Sections 2, 3, and 4 present the perceptual, the attitudinal and the reactive theories.I introduce my argument in favor of the reactive theory in Section 5.In Section 6, I consider and reject one reply to that argument.
Before we start, let me introduce shortly three concepts that are going to play a central role in contrasting evaluative theories of emotions, namely the concepts of intentional content, of intentional object, and of formal object.

| Intentional objects versus intentional contents
All the theories to be considered here are not only evaluative in the sense of explaining emotions in terms of values, but also they are intentionalist: each equates emotions to episodes having intentional objects or contents.How are these two expressions, "intentional content" and "intentional object" to be understood here?While these are sometimes used interchangeably, they will be distinguished in the following way here.The object of a mental episode-if it has one-is what that episode intentionally refers to, what it is about.The content of a mental episode-if it has one-is the way that episode presents its object.Intentional episodes do not just refer to their intentional object (at least, sometimes): they represent their object in a certain way, they ascribe some properties to the object.Such ascriptions of properties to objects correspond to the content of these episodes.This fairly standard way of drawing the object/content distinction originates in Twardowski (1884), another pupil of Brentano.Alternatively, "content" is often characterized in terms of what fixes the condition of satisfaction of mental episodes.The connection between these two characterizations of intentional content is basically this: if the object indeed has the property that the content ascribes to it, then the mental episode is satisfied (Searle, 1983, elaborates on this insight).
The object/content distinction, so understood, is reflected in the way one speaks about objects and contents: while intentional objects are referred to by nouns or noun phrases, contents are expressed via sentential complements that ascribe some predicate to some subject.These sentential expressions may be that-clauses (Mary saw that Bob laugh) or may belong to the family of so-called "small clauses": aspectual constructions (Mary saw Bob as laughing), naked infinitives (Mary saw Bob laugh), gerundives (Mary saw Bob laughing), and so on. 2In this respect, contents always have a predicative structure.This does not entail, or so I shall assume, that contents are necessarily conceptual.One may ascribe a property to an object without having a concept of that property or of that object.

| Formal objects
Theories of emotions make ample use of the concept of formal objects.I shall here stick to Kenny's (1966, p. 66) original characterization of the concept.Following Kenny, the formal objects of a transitive verb are the properties that the objects of that verb must exemplify.Only what is dirty can be cleaned, only what is coloured can be seen, only what is future can be expected, only what seem attainable can be intended… While the relation between an action or a mental state and its particular or intentional object is contingent-other things than this carpet can be dried, other persons than Bob can be killed, other things than this bird can be heard-the relation between that action or mental state and its formal object is necessary: no thing which is not wet can be dried, no being which is not living can be killed, nothing that is not a sound or does not emit sound can be heard.What then are the formal objects of emotions, if any? Contrary to what is sometimes said, the formal objects of emotions cannot be values: innocuous things can be-and often are-feared.A more promising proposal, also due to Kenny, is that the formal object of emotions are apparent values: only what is apparently dangerous can be feared, only what is apparently unjust can be an object of indignation, only what is apparently shameful can be an object of shame.I shall call this the "formal object thesis": Formal object thesis¼ df if x is the intentional object of some emotion, then necessarily, x appears valuable: Kenny describes this necessity as logical; we would rather describe it today as metaphysical, as such claims are not logical laws.We shall see that the formal object thesis, although popular, is not accepted by all evaluative theories of emotions.Let us now turn to contrast and assess these theories.

| THE PERCEPTUAL THEORY
Among early phenomenologists, the perceptual theory of emotions was introduced by Meinong.Tappolet (2000Tappolet ( , 2016) ) has been one of its most prominent contemporary defenders.I shall here focus on her version of the theory.
The perceptual theory is typically introduced by saying that emotions are perceptions of values.This catchy phrase is however misleading in two ways-as perceptual theorists recognize.First, even if perceptual theorists are keen on stressing the analogy between perceptions and emotions, there is one feature of perceptions which they give up right away, namely factivity.If Paul hears the tiger growling, then the tiger is indeed growling (if it is not, then Paul does not hear it growling, but seem to hear it growling).Emotions, contrary to perceptions, are not factive (Tappolet, 2016, p. 15): one may experience the tiger as dangerous while it is not.It is in fact thanks to such axiological illusions that the perceptual theory can account for incorrect emotions.
The second reason why "emotions are perceptions of values" is a misleading motto is that, according to the perceptual theory, emotions are not perceptions of values, but perceptions of objects as being valuable.A perception of value is an objectual, non-propositional perception, such as seeing goodness or feeling beauty.What the perceptual theorist maintains is not that emotions are perceptions of value in abstracto, but rather they are perceptions of objects as having values.Despising somebody is not seeing despicability but seeing her as despicable."-her as (being) despicable" is not objectual, but propositional (or proto-propositional, as some have it): it predicates value of an object.Emotions, according to perceptual theories, are not objectual attitudes directed at values, but contentful attitudes representing an object as exemplifying a value.Otherwise put, all emotions, according to these theories, have an evaluative content.Here is Tappolet: According to the Perceptual Theory, emotions are, in essence, perceptual experiences of evaluative properties.[...] On this account, emotions are claimed to have representational content.They represent their object as having specific evaluative properties (Tappolet, 2016, p. 15).
In sum, instead of speaking of perceptions of values, the perceptual theory is better phrased as the view that emotions are perceptual experiences of their objects as being valuable in some way."Perceptual experiences" are understood as being non-factive, unlike perceptions.
Perceptual theory¼ df emotions are perceptual experiences of objects as being valuable in some way: For example, to be ashamed of one's nose is to experience one's nose as being shameful, to enjoy a book is to experience it as enjoyable, and so on.
One might worry that insisting that the content of emotions has a propositional form will blur the distinction between perceptual and judgment theories.This is not so.What distinguishes the judgment theory from the perceptual theory is not whether or not there is predicative structure in the content of the emotions-intentional contents are by definition predicative, see introduction-but rather whether that predicative content is non-conceptual.On the perceptual theory, positive emotions represent their object as good non-conceptually, whereas on judgment theories emotions represent that their object is good conceptually.Consequently, on perceptual theories, to experience the tiger as being dangerous, my pleasure as being guilty, or Reinach's A Priori Foundations of the Civil Law as being admirable, I do not need to master the concepts of danger, guilt, or admirability.
At this point, one may wonder what it is exactly that we experience when we experience an object as being good.We do not just experience the object, nor do we just experience goodness, nor do we experience a proposition.I think the best proposal is that such experiences represent evaluative states of affairs.By an evaluative state of affairs, I mean a complex of the form [x exemplifies some value], which may or may not obtain, and where the value property is either a universal or a trope.The perceptual theorist may wish to avoid evaluative states of affairs, but it is neither clear why he should, nor whether he could.Some may think that appealing to value-tropes will do the trick: to fear the tiger is to perceive its danger.But this will not work.Even if the danger I experience is the danger of this tiger only, to fear the tiger I need to perceive that danger-trope as inhering in that tiger.And this-a trope inhering in a substance-is a state of affairs, which can only be expressed via a proposition. 3  It is generally agreed that emotions have correctness conditions (admiring Paul may be correct or incorrect), that emotions have cognitive bases (admiring Paul requires having some access to him) and that emotions have formal objects (admiring Paul entails finding Paul admirable).Here is how the perceptual theory accounts for those three key features of emotions.

| Correctness
Emotions can be correct or incorrect: fearing an aggressive and approaching tiger is correct, fearing a dead tiger is incorrect.On the perceptual theory, an emotion is correct if and only if it (or its content) is veridical.My fear of the tiger-experiencing the tiger as being dangerous-is correct if and only if that experience is not illusory, that is, if the lion is indeed as it seems to me: dangerous (Tappolet, 2016, p. 16, 20).Otherwise put, the correctness conditions of emotions boil down to the truth-conditions of their content.Equivalently, an emotion is correct if and only if the evaluative states of affairs it represents indeed obtains.

| Cognitive basis
On top of the factivity issue mentioned above, there is a second disanalogy between ordinary perception and emotions that perceptual theorists acknowledge.While ordinary perceptions do not essentially depend on other perceptions, perceptual experiences of values depend on ("presuppose" in Meinong's terms) non-axiological perceptions.To perceive the tiger as dangerous, I need to perceive the tiger and its relevant natural properties-its having sharp teeth, its looking at me, its growling.These non-axiological perceptions constitute the so-called cognitive basis of emotions.So, emotions are perceptual experiences which require other perceptual experiences to take place.

| Formal object
The perceptual theory is not only compatible with the formal object thesis-to remind, the view that intentional objects of emotions appear valuable-it entails it: the intentional objects of emotions are presented as valuable because emotions just are presentations of their intentional objects as valuable.
Summing up, the perceptual theory endorses these five claims: 1. Emotions are perceptual experiences of objects as exemplifying values.
2. Emotions are correct if and only if their object exemplifies the relevant value.
3. Emotions are correct if and only if their content is true (i.e., if the evaluative state of affairs they represent obtains).
4. Emotions depend on a cognitive basis that provides them with their objects and the relevant natural properties of their objects.
5. The formal objects of emotions are apparent values (formal object thesis).

| THE ATTITUDINAL THEORY
The attitudinal theory refuses to reduce emotions to other mental states or combinations thereof, such as beliefs, desires, judgments, or perceptual experiences.It takes emotions to be irreducible primitive attitudes.What distinguishes emotions from other mental attitudes is that emotions are evaluative attitudes.
Attitudinal theory¼ df emotions are sui generis evaluative attitudes: Deonna andTeroni (2012a, 2015) are the main contemporary defenders of the attitudinal theory, and Brentano (1995) is its chief ancestor.The key distinction between the perceptual theory and the attitudinal theory is that the latter refuses to locate the evaluative aspect of emotions in their content.Emotions do not represent their objects as exemplifying values.Instead, the evaluative aspect of emotions is claimed by attitudinal theorists to be a feature of their intentional modes.The content/mode distinction can be easily introduced through examples.Seeing the sky as blue, seeing Julie as beautiful, seeing the tiger approaching differ in their intentional objects as well in their contents (attitudinal theorists typically do not dwell on the distinction), but not in their attitudes or modes.By contrast, desiring that the tiger goes away, hoping that the tiger goes away and rejoicing that the tiger goes away differ in their attitudes or modes but neither in their object nor in their content.
For a content to be evaluative, as we saw, is for it to be of the form: x is valuable, that is, it is for it to ascribe a value to an object.But what does it mean to claim that a mode is evaluative?The general idea is that emotions do not represent their objects as exemplifying a value but represent their object evaluatively.That is, it is the emotion itself that is evaluative, not its content.This idea was first introduced by Brentano: I do not believe that anyone will understand me to mean that phenomena belonging to this class [feeling and will] are cognitive acts by which we perceive the goodness or badness, value or disvalue of certain objects… this would be a complete misunderstanding of my real meaning.…If we say that every affirmative judgment is an act of taking something to be true, and every negative judgment an act of taking something to be false, this does not mean that the former consists in predicating truth of what is taken to be true and the latter in predicating falsity of what is taken to be false.Our previous discussions have shown, rather, that what the expressions denote is a particular kind of intentional reception of an object, a distinctive kind of mental reference to a content of consciousness.[…] Similarly, then, the expressions which we used in an analogous manner, "to be agreeable as good," and "to be disagreeable as bad," do not mean that in the phenomena of this class goodness is ascribed to something which is agreeable as good, or badness to something which is disagreeable as bad.
Rather, they, too, denote a distinctive way in which the mental act refers to a content (Brentano, 1995, p. 240).
(Note that Brentano uses here "content" to denote what we here call "object"-in general Brentano does not seem to distinguish between content and object).But in which sense are modes evaluative if they do not represent values?What renders certain modes of reference, but not others, evaluative?A certain connection between evaluative modes and values is required, but what is it?Here are three compatible answers.
1. Brentano's idea was to endorse a form of a buck-passing account of value (and truth): something has positive value in virtue of being worthy of being loved, something is true in virtue of having to be affirmed.On this proposal emotions are evaluative in the sense that they figure in the analysans of values, by contrast to other mental modes.Kriegel (2018, chap. 7) as an interpretation of Brentano.It consists in parsing "S experiences x as good" in a non-standard way: "as good" is not a predicate that attaches to x, but an adverb that modifies the experiencing.Hyphens help in phrasing the proposal: emotions represent-as-valuable [their object].For perceptual theories, to fear the lion is to experience the lion as dangerous.For attitudinal theories, so interpreted, to fear the lion is to experience-as-dangerous the lion.

A second way of understanding what makes an intentional mode evaluative is advanced by
3. The third way of understanding the evaluativity of evaluative modes, endorsed by Deonna and Teroni, appeals to correctness conditions.On this approach, an intentional mode is evaluative because, although it does not represent its object as valuable, it is correct if and only if its object is valuable.Thus, fear is an evaluative mode because it lies in the nature of fear that it is correct to fear dangerous objects (but not to represent its object as dangerous).
It is because one adopts a specific emotional attitude towards a given object that the corresponding evaluative property becomes relevant for determining whether the emotion is correct or not.It is for instance because Julianne takes the atitude of fear towards the dog that its dangerousness features in the correctness conditions of her mental state (Deonna & Teroni, 2012a, p. 77).
Deonna and Teroni offer the following helpful analogy, quite in line with Brentano's view of judgment: a belief is correct if and only if it is true.That does not entail that to believe that p is to represent p as being true.By parity, fear is correct if and only if its object is dangerous.That does not entail that to fear an object is to represent it as dangerous.
In all three readings, the attitudinal theorist rejects the perceptualist's claim that an emotion has an evaluative content.How then does the attitudinal theory accounts for the correctness, cognitive basis and formal objects of emotions?

| Correctness
According to the perceptual theory, the correctness of an emotion and the truth of its content go hand in hand.By contrast, on the attitudinal theory, correctness and truth part ways.A first reason for this is that objectual emotions such as admiring Paul or being ashamed of one's nose, appears to lack any content.While perceptual theory analyses such objectual emotions in terms of perceptual state with representational content, such a propositional paraphrase of objectual emotions is alien to the attitudinal theory, which refuses to analyse emotions.Trivially, if some emotions are about objects and lack content-more on this in the next section-the correctness of these emotions cannot consist in the truth of their content.But even in the case of emotions that have content the truth of the emotion's content and the correctness of the emotion must be distinct under the attitudinal theory.This is because the content of the emotions, when there is one, is non-evaluative for the attitudinal theories.Paul regrets that Julie did not come.This emotion has a content which can be true or false.But plainly, the truth of that non-evaluative content does not entail that the emotion is correct.It might be incorrect for Paul to regret that Julie did not come even if it is true that Julie did not come.What makes Paul's emotion correct is not that Julie did not come, but that Julie's not coming is regrettable.Crucially, being regrettable is not part of the content of the emotion.So what makes an emotion correct, on the attitudinal theory, is never that its content is true, either because it has no content, or because that content is non-evaluative.

| Cognitive basis
Like the perceptual theory, the attitudinal theory agrees that emotions have a cognitive basis: to be proud of one's new haircut, one needs first to be presented with it.

| Formal objects
Interestingly, the attitudinal theory is committed to denying the formal object thesis, on pain of collapsing into the reactive theory.As we shall see, the key difference between the reactive and the attitudinal theory is that the attitudinal theory dispenses with any apprehension of value prior to emotions. 4Thus, Deonna and Teroni consider that one chief advantage of the attitudinal theory over both the perceptual and the reactive theories is that it avoids a "mysterious form of non-conceptual acquaintance with evaluative properties" (Deonna & Teroni, 2015).The attitudinal theorist's rejection of the formal object thesis is easily missed as Deonna and Teroni use "formal object" in a sense distinct from Kenny's original proposal.Thus, they maintain that values are formal objects of emotions (Deonna & Teroni, 2012a, pp. 76-77), which may sounds like an endorsement of the formal object thesis.However, as we saw in the introduction, if the formal object of fear is the property F such that only what is F can be feared, then F cannot be the axiological property of being dangerous (if it were, fear could never be incorrect), but rather the psychological property of seeming or appearing dangerous.What Deonna and Teroni call the formal object of the emotions, is instead what makes the emotions correct, namely an axiological property which, ex hypothesis, is neither presented in, nor prior to, the emotion.Terminological quibbles aside, the idea that only what is presented as valuable can be an object of emotions is an idea that the attitudinal theory constitutively rejects.
Summing up, the core claims of the attitudinal theory are: 1. Emotions are sui generis evaluative attitudes that do not (typically) represent values.
2. Emotions are correct if and only if their object exemplifies the relevant value.
3. The correctness of an emotion is not equivalent to the truth of its content.4. Emotions depend on a cognitive basis that provides them with their objects and the relevant natural properties of these objects.
5. Apparent values are not formal objects of emotions (rejection of the formal object thesis).

| THE REACTIVE THEORY
Reactive theories have recently been revived by Mulligan (2007Mulligan ( , 2009aMulligan ( , 2009bMulligan ( , 2010aMulligan ( , 2010b)), Müller (2017aMüller ( , 2017b) ) on the basis of the work of early phenomenologists such as Reinach (1989), von Hildebrand (1916, 1922), Scheler (1973) (see Vendrell Ferran, 2008, 210 sqq. andon Reinach Smith, 2013, 141 sqq.).According to the version of the reactive theory endorsed here, an emotion is an attitude that arises in reaction to some object or state of affairs being apparently valuable.Indignation toward an action arises in reaction to that action being seemingly unjust; fear is an answer to the apparent danger of an object or fact.
Reactive theories¼ df emotions toward some objects are sui generis attitudes which arise in reaction to this object being presented as valuable:

| Formal object
The reactive theory agrees with the perceptual theory that, each time we have an emotion, we represent its object as valuable.Together with the perceptual theory, and contra the attitudinal theory, the reactive theory entails the formal object thesis.But while the perceptual theory equates the emotion with the (non-conceptual) presentation of an evaluative state of affairs, the reactive theory maintains that the emotion is distinct, albeit necessarily accompanied by, such an evaluative presentation.
A first consequence is that the time span of an emotion must be included within the time span of the underlying evaluative presentation.The evaluative presentation may last longer than the emotion, it might start before it, and continue after it, but at no point in time can an emotion exist without its object being presented as valuable at that same time.
Second, the view that it is not the emotion itself that presents its object as valuable, but the evaluative presentation it relies on, allows the reactive theory to accept at once objectual emotions (contra perceptual theories) and the formal object thesis (contra attitudinal theories).Let me explain the point about objectual emotions.Suppose that Mary fears the tiger.Mary's fear has an object-the tiger; but does Mary's fear also have a content?Does her fear ascribe some property to the tiger?Propositionalism (to take up Montague's, 2007, label) answers positively.Although Mary's fear of the tiger is initially reported as bearing only on an object, the fear is in fact, so propositionalism has it, a propositional attitude that represents that object in a certain way.For instance, Mary would not, strictly speaking, fear the tiger, but would fear that the tiger attacks her.If propositionalism is true, then all attitudes that have an object also have a content (in the sense retained here), that is, all attitudes ascribe some feature to their objects.Anti-propositionalism, by contrast, maintains that objectual attitudes should be taken at face value: Mary's fear of the tiger bear just on an object, without ascribing any properties to that object (see Montague, 2007 for a convincing defense and further references).
The reactive theory of emotions embraces anti-propositionalism about objectual emotions: when Mary fears the tiger, her fear does not represent the tiger as exemplifying any property (by contrast to what the perceptual theory maintains, for instance).However, the reactive theory also entails that whenever one fears an object, there is another mental episode that ascribes a value-property to that object (by contrast to what the attitudinal theory maintains).Thus, when Mary fears the tiger, her fear has an object, but no content; but her fear arises in reaction to another mental episode which ascribes dangerousness to the tiger.There are purely objectual emotions, but these depend on presentations having evaluative contents.Propositionalism is false of emotions in general, but true of their evaluative bases.

| Evaluative presentations without emotions
Evaluative presentations are necessary but not sufficient to emotions.On a gloomy day, one might grasp the beauty of a piece of music, get the funny side of a situation, or experience the nice balance of a wine and yet remained unmoved.Our mood, our personality, and other physical and psychological dispositions and contextual features may lead us to have different reactions to a same presented evaluative state of affairs.Thus, writes Reinach, "The meanness of an action can be felt with the same degree of clarity by two men, while their emotional reactions to that state can be entirely different" (Reinach, 1989, vol. 1, p. 296).The leeway for emotional answers is however limited: it is impossible, on the reactive theory, to fear an object because it seems admirable.The only possible reaction to apparent admirability is admiration.It is however possible to remain cold: not to emotionally react to what appears admirable.

| Emotions as attitudes
What are then, if not evaluative presentations, as per the perceptual theory?Here the reactive theory sides with the attitudinal theory: emotions are sui generis attitudes, which should not be analyzed further.Besides, the reactive theory agrees that emotions are evaluative attitudes in Deonna and Teroni's sense: in virtue of their nature, emotions are correct if their object is valuable in the relevant way.However, the reactive theory parts ways with the attitudinal theory on the following key point: while for the attitudinal theory emotions only have a cognitive basis, which presents their object and their relevant natural properties, according to the reactive theory emotions also have an evaluative basis, which presents their object as valuable in the relevant way.
Thus, the reactive theory takes on board the two main elements of the perceptual and attitudinal theories: value perceptions and evaluative attitudes.To reconcile them, it denies that emotions are identical to experiences of evaluative states of affairs while maintaining that such perceptions are necessary to emotions.
Two further elements of the reactive theory call for clarification: the relation of reaction and the idea of an object being presented as valuable.

| Reaction
It is common in ordinary language and in the scientific literature to describe emotions as responses, answers, or reactions.But what is it to affectively react to something being apparently valuable?Emotions are reactions to apparent evaluative states of affairs, first, in the sense that while emotions are distinct from presentations of objects as valuable, it is impossible to have an emotion without its object being presented as valuable in the relevant way (the formal object thesis).But this does not suffice to make emotions reactions to presented evaluative state of affairs.
Burning x is distinct from, and necessitates, x's being inflammable; but burning x is not a reaction to x's being inflammable.What should be added to something being distinct from, and modally dependent on, something else in order for it to be a reaction to it?To react to an object being apparently valuable, I submit, is, furthermore, to adopt an emotional attitude towards that object for that very reason.To fear the tiger in reaction to its being apparently dangerous is to adopt that attitude for that very reason.Mary reacts with admiration to Paul's being admirable, as he appears to her, if Mary admires Paul for the reason that Paul appears admirable to her.

| Evaluative presentation
The reactive theory is sometimes presented as claiming that emotions are reactions to perceptions of values.I believe this is inaccurate, for two reasons.First, in the same way as the perceptual theory does not equate emotions with perception of values-but with perceptions of x as being valuable-the reactive theory, strictly speaking, claims that emotions are reactions not to values, but to objects being apparently valuable.Second and importantly, emotions, I submit, should be seen as reactions not to evaluative experiences but to the contents of such experiences.I fear the tiger for the reason that it is apparently dangerous, not for the reason I experience it as dangerous: there is nothing fearsome about this.The decision makes me indignant because it appears unjust, not because I see it as unjust; again, here is nothing bad about this.Presented or apparent evaluative states of affairs, rather than presentations or appearances of evaluative states of affairs, are what we react to (Dancy, 2000, chaps. 5 and 6).Note that "apparent" or "presented," as they are used here, neither mean nor implicate "illusory," "unreal," or "merely apparent."If the tiger is apparently dangerous, then it may or may not actually be dangerous.When the apparent state of affairs obtains, the evaluative presentation is veridical; when the state of affairs does not obtain, the presentation is illusory.

| Varieties of evaluative presentations
Are evaluative presentations all as per the perceptual theory, or should the reactive theory include other forms of evaluative presentations?Admittedly, one chief way for evaluative states of affairs to be apparent is for them to be experienced: non-conceptually grasped, perceived, or felt.The idea that evaluative states of affairs can be perceived, felt, experienced or non-conceptually represented, which lies at the heart of the perceptual theories, is taken on board by the reactive theory (which is therefore, like the perceptual theory, open to the worry that such "graspings of value" would be "mysterious").But the reactive theory, as here defended, is less restrictive than the perceptual theory in that it admits other forms of evaluative presentation.Evaluative perceptions are meant to be (a) non-conceptual-in the sense that they represent x as valuable without mobilizing axiological concepts; (b) assertive-in the sense that they present their content as obtaining; (c) present-directed-in the sense that they present their content as present.Evaluative bases for emotions may fail to have some or all of these features.Thus evaluative presentations may include: seeing the countryside as bucolic, experiencing the Alps as beautiful, anticipating that the dinner will be boring, expecting Paul to be a coward, remembering a joke as a funny one, grasping the elegance of a proof, believing that my neighbor is a murderer, imagining that a dish is disgusting, and so on.Thus construed, the reactive theory can accommodate emotions prompted by evaluative beliefs or judgments (e.g., the fear trigged by the belief that nuclear weapons are dangerous and disseminated in the world, to take up Dokic & Lemaire, 2015's example), intellectual evaluative intuition (Bengson, 2015;Huemer, 2005), evaluative rememberings, evaluative anticipations, and so on.It can also accommodate cases in which the evaluative content, albeit nonconceptually represented, is not asserted, as in evaluative imaginations, evaluative suppositions, or evaluative considerations of a thought.Imagining a melody may spark aesthetic enjoyment. 5

| The object of the emotions and the content their evaluative bases
Evaluative presentations have propositional content: they present an object as exemplifying a value.That content-an evaluative state of affairs-is however necessarily distinct from the content of the emotions that arises in reaction to it.The object of the emotions is not the evaluative state of affairs represented in the evaluative presentation, but it is only the value-bearer in that content.More precisely, when the value-bearer represented in the evaluative presentation is an object-the tiger is dangerous-the ensuing emotion is objectual-to fear the tiger.When the value-bearer represented in the evaluative presentation is a state of affairs-it is dangerous that the tiger is in the room-the ensuing emotions is propositional-to fear that the tiger is in the room.Fearing a tiger is not the same as fearing that there is a tiger in the room; on the reactive theory, both have distinct evaluative bases.More generally, while evaluative bases are always directed at evaluative states of affairs, our emotions are rarely directed at evaluative states of affairs: first because some emotions are objectual and second because even among emotions that are directed at states of affairs, many are not directed at evaluative states of affairs: we may be ashamed not to be courageous enough, but we fear that there is a tiger in the room, we do not fear that it is dangerous that there is a tiger in the room.

| Cognitive and evaluative basis
For the reactive theory, emotions not only have a cognitive basis (as per the perceptual and attitudinal theories), but also an evaluative basis.Depending on the cases, as we just saw, that evaluative basis may be conceptual or nonconceptual, may represent the evaluative state of affairs as obtaining or not, as obtaining now, in the past, in the future, eternally, and so on.

| Correctness
The perceptual theory equates the correctness of emotions with their veridicality.In line with the attitudinal theory, the theory denies that correctness of emotions boils down to the truth of their content, and more generally that emotions are always truth-apt (objectual emotions are not).Yet, together with the perceptual theory, it insists that an emotion is correct if and only if some content is true, namely the content of the evaluative presentation on which the emotion depends.
This proposal is open to the following objection. 6In case of Schadenfreude, one enjoys somebody else's misfortune.Here we seem to have an underlying evaluative presentation to the effect that the person's situation is bad, which is veridical.Yet, the emotion is intuitively incorrect.So, the objection goes, the veridicality of the evaluative basis and the correctness of the emotions may come apart: sometimes emotions are incorrect because they are reactions to non-veridical evaluative basis, sometimes they are incorrect because they are a wrong reaction to veridical evaluative basis.The answer, I submit, is that Schandenfreude is not a reaction to the apparent fact that a given person is in a bad situation, but to the fact that it is valuable to us that that person is in a bad situation.That is, the reason why we enjoy somebody else's misfortune, is that her misfortune seems good to us in some way.This is in accordance with the formal object thesis: when we enjoy somebody else's misfortune, her misfortune is presented as good for us.If this is false-that is, if the person's misfortune is in fact not good for us-then the episode of Schadenfreude is incorrect.If, on the other hand, the person's misfortune is good for us, then the episode of Schadenfreude is correct (which is not to say, nor to deny, that it is right or ethical; see D 'Arms & Jacobson, 2000).
Summing up, the core claims of the attitudinal theory are: 1. Emotions are sui generis evaluative attitudes that do not (typically) represent values.
2. Emotions are correct if and only if their object exemplifies the relevant value.
3. The correctness of an emotion is not equivalent to the truth of its content (if it has one) but is equivalent to the truth of the content of its evaluative basis.
4. Emotions depend on a cognitive basis that provides them with their objects and the relevant natural properties of these objects.
5. Emotions depend on an evaluative basis that provides them with their reasons.
6.The formal objects of emotions are apparent values (formal object thesis).

| EVALUATIVE REASONS FOR EMOTIONS
The perceptual, attitudinal, and reactive theories share significant hypotheses: all agree that emotions should be accounted for in evaluative terms; all agree that emotions are intentional; all agree that emotions are correct if and only if their object exemplifies the relevant value.The bone of contention between them concerns the evaluative presentations: the perceptual theory equates emotions with evaluative presentations; the attitudinal theory denies that emotions are evaluative presentations and that emotions depend on evaluative presentations; the reactive theory denies that emotions are evaluative presentations but maintain that emotions depend on evaluative presentations.
My argument in favor of the reactive theory is this.The reason why we fear the tiger is that it appears dangerous; we admire Mary because she seems admirable.Neither the perceptual theory nor the attitudinal theory can accommodate such evaluative reasons for emotions.Only the reactive theory is in a position to do so, as it distinguishes emotions from evaluative presentations without getting rid of evaluative presentations.More precisely: (P1) Emotions are states for which we have reasons.
(P2) The perceptual theory entails that reasons for emotions, if any, are non-evaluative.
(P3) The theory entails that reasons for emotions, if any, are non-evaluative.
(P4) The reactive theory entails that reasons for emotions are evaluative.
(C) The perceptual and attitudinal theories are false, only the reactive theory may be true.
I now turn to explaining and defending these premises.

| P1: Emotions are states for which we have reasons
We often enquire about the reasons why people admire, enjoy, fear or despise some object-the motivational reasons for emotions.We also often wonder about whether people should admire, enjoy, fear, or despise some objectthe normative reasons for emotions.When everything goes well, the reason why one has an emotion is a good one: a reason why one should have that emotion.That there are reasons for emotions does not entail that emotions are actions-like: admittedly, we have no direct control over our emotions, we cannot fear, admire, or rejoice at will.But this does not prevent us from having justifications for our emotional reactions.Emotions are, in this respect, on a par with beliefs, and contrast with perceptions.One can meaningfully ask "Why/for what reason are you afraid of the tiger?," "On what grounds do you admire Paul?," in the same way as we can inquire about reasons for beliefs.But there are no reasons for seeing a tree, smelling a rose or feeling a pain, not even bad reasons.One may indeed wonder about the causes of our perceptions, in the very same way that one may wonder about the causes of our emotions.It might be the case that Mary had an outburst of shyness because she is very shy; that Bob was amused because he smoked marijuana; or that Paul fears the snow because he was raised in Kenya.But in the case of beliefs and emotions we may also provide non-causal, justificatory, explanations: Bob was amused by the joke because (=for the reason that) it is funny, he finds; Paul fears nuclear plants because they are dangerous, he thinks; Elisabeth is indignant about the decision because it is unjust, she feels. 7The "because" in these sentences is the "because of reason," in contrast to the "causal because" (Mulligan, 2007).There are reasons for intentions, beliefs, desires, and emotions; but there are no reason for perception, sensation, knowledge, and feelings.P1 has however recently been challenged by Maguire (2018), who denies that there are reasons for emotions.
Could attitudinal and perceptual theorists rely on Maguire's rejection of P1 to resist the above argument?I think not.
To begin with, perceptual and attitudinal theorists are not willing to do so: Tappolet as well as Deonna and Teroni accept P1.But other perceptual and attitudinal theorists may still adopt that move.So I need to show that Maguire's arguments that there are no reasons for emotions are inconclusive.Maguire questions the claim that there are reasons for emotions on the grounds that alleged "reasons" for emotions are disanalogous to reasons for action.The chief disanalogy he pinpoints is this: while one action can be supported by several reasons, which are weighted against each other, this is not the case for emotions: when several "reasons" for affective reactions arise, they normatively support different affective reactions.The difference between "normative supports" that can compete and those that cannot is so crucial, Maguire claims, that we had better keep the term "reason" for the former normative supports only.Since he holds that reasons are "essentially contributory," he maintains that normative supports for emotions cannot be reasons.
Maguire has I think spotted an important and neglected difference: actions can be supported by several competing reasons, emotions cannot.But I believe he draws the wrong conclusion from this acute observation.Maguire locates the source of the disanalogy in an essential difference between normative supports for emotions and normative support for actions.I believe on the contrary that the disanalogy stems from some essential differences between emotions and actions.
Let me introduce this idea by a hopefully illustrative metaphor.The composition of forces is often appealed to in order to shed light on the weighting normative supports. 8In the same way that three-dimensional bodies can be acted on by several forces, actions can be supported by different reasons.Reasons are to actions what forces are to bodies.Consider now ghodies.Ghodies are extended entities which are partly ghostly, partly body-like.Like ghosts, ghodies are penetrable everywhere without resistance, except at one particular point of their surface.At that single point (and only at that point), if one presses them in a specific direction and at a determinate degree, one will change their motions-quite like bodies.Thus, the only way to mechanically change the motion of a ghody is to press it at a unique definite point, in a unique definite direction and at unique definite degree.Pressing it elsewhere, or too lightly or too strongly, or not at the right angle, will result in compenetrating that ghody without affecting its motion.As a consequence, it is impossible for a ghody to be acted on by several forces. 9Now, that ghodies cannot be acted on by several forces is not due to the nature of the forces acting on them, but to the nature of ghodies.Ghodies cannot be acted on by several forces, because ghodies are highly selective regarding the forces that can act on them.It would be very odd to claim that the "forces" that act on ghodies, because they cannot compose with each other, are in fact not forces but constitute a distinct kind of "dynamical support," essentially distinct from the forces that affect bodies.
Emotions, I submit, are like ghodies: highly selective about the kinds of reasons that can move them.They can only be normatively supported one way.Irrelevant reasons leave them unaffected.That several reasons cannot support the same emotion is due to the nature of emotions, not to the specific features of normative supports for emotions.Likewise, that several reasons can support the same action, is due to the nature of action, which, like solid three-dimensional bodies, are sensitive to many different influences. 10This hypothesis accommodates Maguire's data while being compatible with the common view that emotions are supported by reasons.
Finally, note that Maguire does not deny that emotions have normative supports.He only questions the idea that such normative supports belong to the category of reasons.The reactive theorist endorsing Maguire's argument may perhaps rephrase P1 in terms of "normative supports for emotions" rather than "reasons for emotions."This sounds less natural but is perhaps no less effective against perceptual and attitudinal accounts.

| P2:
The perceptual theory entails that reasons for emotions, if any, are nonevaluative It has been objected to perceptual theories that they cannot account for reasons or justification for emotions, because perceptions lack reasons and are not justified (Brady, 2011).The present worry is compatible but distinct.The point is that, if perceptual theories can provide reasons for emotions, these reasons can only be nonevaluative.The argument for this is relatively straightforward: according to perceptual theories an emotion is the perception of an object as being valuable.The reason why one perceives an object as valuable cannot be that that object is perceived as valuable by one.Explanation in general, and explanation by reasons in particular, Thus, while it is true that emotions differ from sensory experiences with respect to justification, this has more to do with the nature of values than with the nature of emotions (Tappolet, 2016, p. 40).
Perceptual theories, to the extent that they can vindicate reasons for emotions, have to equate those reasons with the non-evaluative supervenience basis of the values represented by the emotions.
This worry is even more pressing in the case of more complex or less basic emotions such as shame, admiration, pride, where the supervenience basis of the relevant values also includes historical, personal, social, arguably even axiological factors.It culminates perhaps in cases of aesthetic pleasures.As emphasized by Sibley (1959Sibley ( , 1965)), determining the non-evaluative properties sufficient for the exemplification of thick aesthetic values is typically infeasible.But if the reasons for enjoying Rameau's music lie in the full list of the non-evaluative value-making properties-which is going to include intrinsic and extrinsic properties of the piece of music, perhaps as well as properties of the listener, we would never be justified in enjoying it.
The worry vanishes, on the other hand, if reasons for emotions are identified with evaluative state of affairs rather than with their supervenience basis.Our reason for fearing the lion is that it is dangerous, not the complex supervenience basis of that dangerousness.Better, this solution might even provide the beginning of an answer to Deonna and Teroni's objection, to the effect that experiences of values would be mysterious.For what the remark above suggests is that when trying to get at the complete supervenience basis of a given value exemplification, what we start from is the knowledge of that value.What strikes us first is the lion being dangerous, the situation being shameful, or the wine being balanced.We thereafter enquire about what exactly makes them so (a point stressed by Scheler, 1973, pp. 17-18; see also Sibley, 1965, pp. 140-141).
The second reason for rejecting the view that reasons for emotions are non-evaluative is that its intuitive appeal relies on a surreptitious reintroduction of evaluative justifiers (I here elaborate on an original insight from Müller, 2017aMüller, , 2017b)).When we say that our reason for fearing a lion is that it has sharp teeth and is approaching fast, we smuggle in a value property, namely the danger of the lion.More generally, the plausibility of claims justifying emotions by appealing to non-evaluative properties depends on an implicature to the effect that the relevant evaluative property is exemplified.That there is an implicature in such cases is suggested by the fact that it can be cancelled.That reasons for emotions are evaluative is shown by the fact that, once this implicature is cancelled, sheer appeal to non-evaluative properties do not justify emotions anymore.Thus the following claims sound clearly odd: • I am afraid of the lion because it has sharp teeth and is approaching fast, which does not make him dangerous in the least.
• I am amused by this situation because nobody expected it.There is nothing funny about its being unexpected.
• I am angry because my bike was stolen.There is nothing wrong with it having been stolen.
• I admire her because she is an alpinist.Being an alpinist does not render her admirable.
• I appreciate that vase because it is very thin-a thinness that confers no aesthetic property onto it.
What such examples make clear is that, when we invoke non-evaluative properties to justify our emotions, it is indeed the implicated value-properties which are doing the justificatory work.Cancel these implicatures and the justification collapses.Note that the reverse does not hold: mentioning a value property implicates in some context some non-evaluative properties.But such implicatures can be cancelled without threatening justification.Thus, "I am afraid of the lion because it is dangerous, but not because of its sharp teeth" is still fine.Natural properties alone are not sufficient to justify emotions, only value properties are.I conclude that the proposed argument is correct.We should endorse the conclusion: reject both the perceptual and attitudinal theories and embrace the reactive theory.

| DEONNA AND TERONI'S REPLY
Deonna and Teroni anticipated the above worry.They grant that the view that reasons for emotions are evaluative has some attraction (Deonna & Teroni, 2012a, p. 97).On top of the intuitive appeal of the proposal, there is arguably a further reason why the attitudinal theorist may not want to drop the idea that values are apprehended.As noted above, the three evaluative theories of emotions considered here agree that emotions are correct if their object exemplifies the relevant value.But while the perceptual and the reactive theories also agree that these values can be accessed, the attitudinal theorist, on the face of it, has to say that there are values out there, that they make our emotions correct, but that we do not have immediate access to them.This sounds dangerously close to view that emotions are made correct by noumenal entities.
To tackle these issues, Deonna and Teroni introduce the relation of "constitution" which, they claim, holds between non-evaluative and evaluative properties.This relation is at once an explanatory relation and a "sort of" token identity relation: To say that an object's exemplification of an evaluative property is constituted by its exemplification of non-evaluative properties is to say that this object exemplifies the former in virtue of or because of its exemplifying the latter: there is a sort of token identity here (Deonna & Teroni, 2012b).This proposal is I believe inconsistent.If the dog is dangerous because of the sharpness of its teeth, its dangerousness cannot be the same thing as the sharpness of its teeth; explanation is irreflexive, identity is reflexive.We cannot have our cake and eat it.But perhaps they do no really have identity in mind, but something close to it.In other places indeed, they eschew talk about identity, and say that values are "no further properties" than then nonevaluative properties that constitute them: if danger is constituted by the instantiation of some non-evaluative properties, there is no further fact of the matter, nothing more to a specific danger than the instantiation of what makes it a danger.In a given context, a dog with big teeth and impulsive behavior constitutes a danger, as the death of a person may constitute a loss, or a specific remark may constitute an offense.An instance of danger, loss, or offensiveness is not a further property alongside those properties that constitute it (Dancy, 1993, p. 75).If awareness of these properties explains why a subject undergoes the relevant emotion, then this is explained by the subject's awareness of an instance of the relevant evaluative property.And this seems sufficient to justify it (Deonna & Teroni, 2012a, p. 97).
The proposal seems to be that although evaluative properties are not identical with non-evaluative properties (on pain of falling back into the previous inconsistency), evaluative properties are nonetheless nothing further than the non-evaluative properties that constitutes them.The view that evaluative properties are distinct from non-evaluative one, but nothing more than them sounds like an oxymoron.In fairness, Deonna and Teroni are here in good company.They take their lead from Dancy (1993), who argues that evaluative properties, although being nothing over and above non-evaluative ones, are nevertheless distinct and explained by non-evaluative properties.Similar claims have been advanced in the context of grounding theory, where full grounding has been held to amount to reduction and nothing over-aboveness: it is natural [in cases of grounding] to say that the explanans or explanantia are constitutive of the explanandum, or that the explanandum's holding consists in nothing more than the obtaining of the explanans or explanantia (Fine, 2012, p. 39, see also Fine, 2001, p. 15;and Rosen, 2010).
I confess to being unable to make sense of the idea that x is nothing more than y while being distinct from y.I also fail to make sense of the idea that two facts are distinct but that the obtaining of the one is nothing more than the obtaining of the other.But to avoid the shifty notions of constitution and reduction, the present worry may be rephrased in terms of identity alone.Either instances of value properties are identical with their supervenience basis, or they are not.If they are not, the attitudinal theory entails that justifiers for emotions are non-evaluative, as argued above (and as Deonna and Teroni seem to recognize in some places).If they are, the attitudinal theory can indeed welcome evaluative justifiers, but it then collapses into a version of the reactive theory, namely a reactive theory that embraces naturalism about values.
I have argued that reactive theories of emotions are far better than perceptual and attitudinal theories when it comes to the justification of emotions: only the reactive theory can accommodate the fact that values constitute reasons for our responses.This argument, if correct, does not yet establish the truth of the reactive theory of emotions.
First, one may reject evaluative theories of emotions altogether, on the grounds that, for instance, we should explain values in terms of emotions instead of explaining emotions in terms of values.Second, the perceptual or attitudinal theorist may grant the argument above but maintain that, on balance, their theories remain preferable to the reactive theory.The chief difficulty for the reactive theory is to give support to the idea that there are apprehensions of values which are not yet emotions (see Mitchell, 2019).I have not addressed that difficulty here.To many, non-emotional values feelings/perceptions/graspings/apprehensions/intuitions are so weird that they are ready to get rid of the idea that reasons for emotions are evaluative to avoid them.My inclination is the opposite: that reasons for emotions are evaluative constitutes a strong reason to think that evaluative states of affairs can be pre-emotionally apprehended.11ACKNOWLEDGMENT Open access funding provided by Université de Neuchâtel.

ENDNOTES
1 Evaluative theories of emotions can be contrasted with emotional theories of values, which explain values in terms of emotions.If explanation is asymmetric, evaluative theories of emotions and emotional theories of values are incompatible.some given emotions.But the theory entails that this is not necessary and that it is in fact unusual.
5 Aesthetic pleasure is often claimed to be indifferent to the reality of its object: an imaginary landscape may appear bucolic and be enjoyed as such, whether or not it is taken to exist (see Meinong, 1972, sec. 10, see also p. xxi).
6 I am grateful to a referee of this journal for having raised that worry.
7 I am here following Dancy (2000, ch. 5) in using such appositional constructions.The reason for preferring "Bob laugh for the reason that the joke is funny, he finds" to "Bob laugh for the reason that he finds the joke funny," is that the former makes clear that it is the content of the finding-that the joke is funny-, rather than the finding itself that constitutes the reason of Bob's laugh (as argued in subsection 4.5 above) 8 See, for example, Ross (2002, pp. 28-29).
9 I am assuming, quite standardly, that forces are individuated by their point of application, magnitude, and direction. 10Maguire mentions another disanalogy between normative supports for emotions and normative supports for actions: only the later are gradable.I again agree on the difference but would argue again that this difference stems from essential distinctions between actions and emotions rather than from any essential difference between normative supports.
is irreflexive.Consistently, Tappolet does not attempt to provide evaluative reasons for emotions.According to her, emotions have reasons, despite being perceptions, because (part of ) what they are perception of-valuesare essentially supervenient properties.Then, what justifies our emotions is the non-evaluative properties on which the perceived value supervenes: the fact that a dog has [sharp teeth and a short temper] gives you reason to perceive it as fearsome […]