Perceptualists maintain that emotions essentially involve perceptual experiences of value. On this approach, anger might be thought to involve an experience of offense, pride an experience of one’s own achievement, and so on. The perceptual approach has enjoyed significant support in emotion theory (Roberts 2013; Tappolet 2016, inter alia). Theorists have also relied on it in value epistemology (Milona 2016), action theory (Döring 2007), and normative ethics (Stockdale 2017). To be sure, perceptualist theories vary in the details, including important ways that I canvass below. But despite such differences, perceptualists are unified in taking emotions to have evaluative content in much the way that visual, auditory, etc. experiences have empirical content.

At first glance, perceptualism looks like a promising starting point for analyzing emotions. Many philosophers today maintain that emotions are not (mere) bodily sensations; they are evaluations. It was once popular to treat these evaluations as forms of judgment (Solomon 1976; Nussbaum 2004). But many have since migrated from judgmentalism, as it is often called, to perceptualism. One major reason for this trend is simple. When we are overcome with fear, to take a familiar example, we sometimes explicitly judge that what we fear isn’t dangerous. But such cases are not experienced as similar to making contradictory judgments (see D’Arms and Jacobson 2003; Naar 2020). They instead seem more akin to perceptual illusions, whereby things appear other than we believe them to be (Tappolet 2016). So if we accept that emotions are evaluations, then a perceptual model looks like a promising starting point.

However, Julien Deonna and Fabrice Teroni (2012, 2015) forcefully argue that perceptualism isn’t a great starting point after all. While they agree that emotions are evaluative experiences, they maintain that perceptualism goes awry in treating all emotions as being the same type of attitude. This leads perceptualists to distinguish emotion types by virtue of their supposedly differing evaluative content. But, Deonna and Teroni argue, there are several ways in which ordinary, pretheoretical ways of talking and thinking about emotions conflict with emotions being distinguished by their evaluative content, or even having such content at all. I refer to these objections as the Attitudinalist Objections, or jointly as the Attitudinalist Challenge. They maintain instead that the evaluative dimension of an emotion is a feature of the attitude rather than its content; and because this evaluative dimension is different for each emotion, each emotion is a different type of attitude. Their theory is thus a version of attitudinalism, according to which emotions are evaluative attitudes but do not have evaluative content.¹

This paper defends perceptualism in the face of the Attitudinalist Challenge.² I argue that the objections either rely on subtle mistakes about what perceptualism says, or else turn on optional commitments that perceptualists can avoid on independent grounds. Having argued how perceptualists should answer the Attitudinalist Challenge, the paper closes by issuing a challenge of its own for versions of attitudinalism that share perceptualism’s commitment to the view that emotions are evaluative experiences.

1 What Perceptualism Is

Versions of perceptualism have been defended by Cooper (1699), Meinong (1972), Roberts (2013), and Tappolet (2016), among others. The basic view is as follows:

Sabine Döring offers an intuitive illustration with reference to the emotion of indignation:

In experiencing indignation at the harsh punishment of the toddler, it seems to you that the punishment is in fact unjust: your occurrent emotional state puts forward your indignation’s content as correct. This is in analogy to the content of a sense perception. In perceiving that the cat is on the mat, it seems to you that the cat is actually there. (Döring 2007, 377)

Several aspects of perceptualism require clarification. First, the theory speaks of “emotional experiences” because it is only meant as an analysis of occurrent, conscious emotions. For example, while it may be true that Cassandra loves Sasha even when Cassandra is sleeping, “love” here is meant dispositionally. Perceptualism is not about emotions in this sense.

Second, the phrase “essentially involve” is non-committal about a key question, namely whether there are any necessary components of emotion other than non-doxastic, affective representations of value. We can thus distinguish between the following positions:

To illustrate these two positions, consider efforts to analyze emotions commonly begin by listing paradigmatic features of emotional experiences. These include evaluations, bodily feelings, action tendencies, and patterns of attention, among other things (Brady 2018, 10). For example, a hiker who fears a nearby bear can be expected to evaluate the bear as dangerous, experience sensations characteristic of fear, be motivated to avert the threat, and attend to whether the bear really is dangerous and what the escape options might be. Identity perceptualists maintain that emotions are in essence their evaluative dimension, which they take to be a non-doxastic representation of value. By contrast, parthood perceptualists see the evaluative dimension as insufficient on its own. Perhaps, for example, it must be paired with a tendency to act in accordance with that representation. In the case of fear, for example, this might be a tendency to act so as to avoid what is experienced as dangerous.

Perceptualism’s advocates are almost always identity perceptualists.³ This may come as a surprise, given that perceptualism’s close cousin, judgmentalism, does divide into two distinct camps. That is, there are some who think that judgment fully captures the nature of emotion (e.g., Nussbaum 2004), and others who think it must be supplemented (e.g., Green 1992).⁴ As it happens, many of the motivations for perceptualism, including that it can provide a plausible basis for value epistemology and that it can explain how emotions rationalize action, only require parthood perceptualism. Furthermore, perceptualists who are willing to take seriously parthood perceptualism have additional resources for addressing the Attitudinalist Challenge. For if emotions include more than evaluative representations, then they may be distinguished not only by their evaluative content, but also by other features (e.g., action tendencies).⁵ However, because identity perceptualism is the dominant version of the theory, and because the Attitudinalist Challenge is most serious for this version, I focus in what follows on identity perceptualism.

Other key questions for perceptualists concern the relationship between an emotion’s purported evaluative content and its phenomenology. Perceptualists typically view an emotion’s representation of value as inseparable from its affective (felt) dimension. Here is Roberts:

Affect is not something in addition to emotion […] Just as in the visual experience of a house one is appeared to in the way characteristic of house-sightings, so in fear one is appeared to (in feeling) in the way characteristic of threat confrontations (the threat being directed at something one cares about). (Roberts 2013, 47–48)

Others make similar claims about the inseparability of emotional affect/feeling and the representational dimension of emotion (e.g., Döring 2007, 374; Tappolet 2016, 27–28; see also Ballard 2021b, 121).⁶ According to this position, to describe what it is like to have an emotional experience requires reference to value (see Poellner 2016, 270). In experiencing, say, anger, we cannot describe its phenomenology without reference to the property of being wronged. I take perceptualists to be committed to this inseparability of emotional phenomenology and value. Such a position is compatible with different views about the relationship between how an emotion feels and what it represents. For example, on one possible view, the affective aspect of an emotion (or at least part of it) grounds the evaluative representation. This would accord with an increasingly popular approach to perceptual content which grounds such content in the phenomenal character of perceptual experience (see Kriegel 2013). But here I am non-committal about whether the intentionality or phenomenology of emotions is more basic (if either is).⁷

Additional details about how perceptualists should, or at least reasonably can, develop their view will emerge in the course of addressing the Attitudinalist Challenge. In particular, I suggest that perceptualists take up more specific views about the affective, non-doxastic representation of value and how it relates to ordinary sensory experience.

2 The Attitudinal Alternative

The attitudinal theory is an important alternative to perceptualism. While attitudinalists agree that emotions are evaluations, they deny that emotions have evaluative content (e.g., Deonna and Teroni 2012, 2015; Müller 2017). Emotions are taken to be evaluative at the level of attitude.

The basic idea can be illustrated by way of a comparison with belief and truth. A belief that \(P\) has \(P\) as its content. But there’s more to a belief than its content. After all, one can also suppose that \(P\). One major difference between a belief and a supposition with the same content is that the former is in some sense truth-directed. However, a belief that \(P\) doesn’t represent that \(P\) is true, for a belief that \(P\) has different content than a belief that \(P\) is true (see Kriegel (2019b), 10; Ballard (2021a), 852–853). So truth somehow characterizes the very attitude of belief. That is, a belief is a way of taking-as-true some content. According to Deonna and Teroni, matters are similar with emotion, except that values, rather than truth, characterize emotional attitudes. So instead of saying, for instance, that fear represents the property of being dangerous and anger represents the property of being offensive, the attitudinalist says that the attitude of fear is a way of taking-as-dangerous its content and that the attitude of anger is a way of taking-as-offensive its content.

But what is it to take-as-dangerous or take-as-offensive? On the most widely discussed version of attitudinalism, we find another similarity with perceptualism: emotional experiences are a way of experiencing value (Deonna and Teroni 2012, 2015).⁸ And as with perceptualism, when all goes well, these are experiences through which we come to apprehend objects as having certain values. Deonna and Teroni describe these experiences in terms of the form of readiness to act involved in each emotion (cf. Frijda 2007). Here are two helpful illustrations:

Fear of a dog is an experience of the dog as dangerous insofar as it is an experience of one’s body being prepared to forestall its impact (flight, preventive attack, immobility, etc.), an attitude it is correct to have if, and only if, the dog is dangerous. In the same way, anger at a person is an experience of offensiveness insofar as it consists in an experience of one’s body being prepared to retaliate, an attitude that is correct if, and only if, the person is offensive. (Deonna and Teroni 2015, 303; see also Deonna and Teroni 2012, 81)

Since Deonna and Teroni’s theory explains the sense in which emotions are evaluative experiences by appealing to such action tendencies, I refer to this as action tendency attitudinalism.⁹ By maintaining that emotions are ways of experiencing value, one might suppose that action tendency attitudinalists can thereby secure many of the advantages (or at least ambitions) of perceptualism in value epistemology and action theory. I briefly address these matters in the penultimate section.

It is important to note that while Deonna and Teroni’s action tendency attitudinalism is often treated as the representative version of attitudinalism (e.g., Rossi and Tappolet 2019; Ballard 2021a), the theory can take different forms. Attitudinalism as such merely claims that emotions are evaluative at the level of attitude rather than content. Thus an attitudinalist might agree with Deonna and Teroni that emotions are evaluative experiences but resist the idea that this has to do with experiences of action readiness (cf. Kriegel 2019b, 13). I consider below (section 6) why an attitudinalist might favor such an alternative characterization of emotions as evaluative attitudes. Furthermore, it is also consistent with attitudinalism to maintain that emotions aren’t ways of experiencing value at all. For example, Müller (2017) argues that emotions are responses to pre-emotional experiences of value rather than experiences of value themselves; and these responses are such as to be correct in the presence of the relevant value.¹⁰ So attitudinalism is highly flexible. To keep things manageable, however, I limit my discussion to versions of attitudinalism that take emotions to be evaluative experiences and likewise focus the ensuing discussion primarily on action tendency attitudinalism.

3 The First Attitudinalist Objection: Perceptualism as a Bad Start

The first Attitudinalist Objection is simple, at least in outline. It emerges from similarities between how we pretheoretically conceptualize different emotions as compared to attitudes such as belief, desire, perception, etc. Here is how Deonna and Teroni put it:

[R]egarding the different types of emotions as different attitudes and not as one and the same attitude—for example the attitude of judging or that of perceiving—towards different contents is the default position […] Isn’t it natural to understand the contrast between, say, fear, anger and joy as one between different ways the mind is concerned with objects and events? Shouldn’t this contrast be located at the same level as that between desiring, believing and conjecturing and be clearly distinguished from the contrast between believing a given proposition and believing another? (Deonna and Teroni 2015, 296)

The argument can be summed up as follows. The first premise is that when we talk about believing, desiring, perceiving, etc., we are talking about different attitudes. The second premise is that if the foregoing premise is true, then by analogy, when we talk about emotions, including fear, envy, and so on, it is natural to assume that we are also talking about different attitudes. But, the argument continues, perceptualism denies that emotions are distinct attitudes. For according to perceptualism, all emotions are constituted by the same affective attitude. Call this the Perceptualism as a Bad Start Objection. It is easy to see why Deonna and Teroni, building on this objection, maintain that attitudinalism, rather than perceptualism, should be our starting point for theorizing the sense in which emotions are evaluations.

Before considering how the perceptualist might reply, we should consider what it is for something to be an attitude. There are different ways in which one might define such a technical (or quasi-technical) term. But as the passage from Deonna and Teroni above illustrates, they intend for the purposes of this objection a sense of “attitude” inclusive of perceiving (Deonna and Teroni 2015, 296; see also Kriegel 2019a). This makes sense given the present dialectic. The objection isn’t that perceptualists fail to treat emotions as attitudes; it’s that they treat all of them as the same attitude, distinguished only by their contents.¹¹ Furthermore, defenders of perceptualism have recently been explicit that they don’t mean to deny that emotions are attitudes (Rossi and Tappolet 2019, 553). I thus suspect that Deonna and Teroni have in mind a capacious view of attitudes whereby an attitude is “a way of having content” (Siegel 2021). A perceptualist would certainly grant that emotions are attitudes in this sense.

In addressing the Perceptualism as a Bad Start Objection, I focus in particular on experiential ways of having content. By this I mean to refer to ways of having content such that there is something it is like to represent in that way.¹² By focusing on experiential ways of having content, perceptualists can ensure that their response hews close to the surface of our emotional life and so doesn’t lose sight of the intuition driving the objection. I therefore won’t be concerned with sub-personal ways of representing, or with sub-personal processes that give rise to experiences with certain content (cf. Siegel 2021; Kriegel 2019a). To illustrate, suppose that a perceptualist attempts to address the challenge by appealing to distinct neural machinery underlying different emotions (see Tracy and Randles 2011). The various processes by which different emotions arise may lead a perceptualist to say that there are many different emotional attitudes insofar as they involve the functioning of distinct biological capacities. But it seems to me that Deonna and Teroni’s objection doesn’t hinge on the underlying neural architecture of emotion but is rather focused on the surface of how we pretheoretically talk and think about emotions.¹³ By focusing on attitudes as experiential ways of having content, then, we mitigate the risk of missing the point.

How, then, should a perceptualist respond to the objection? The most straightforward reply is already suggested by the core of perceptualism, namely its analogy with sensory experiences. To see why, recall that the objection invites us to have the intuition that just as perception, belief, and desire are all distinct attitudes, so too are the various emotions, including joy, anger, sadness, etc. But there are alternative comparisons that, from a pretheoretical perspective, we might just as easily have made. More specifically, we might have compared emotional experiences and experiences in different sensory modalities, including visual, auditory, tactile, etc. experiences. Here again, the focus is on the sensory experiences themselves, rather than the underlying sub-personal processes.¹⁴ And here too we can ask what makes an experience in one modality experientially, or phenomenologically, distinct from an experience in another modality. One salient difference, of course, concerns the contents of experiences in different modalities. For example, a visual experience has colors as part of its content while an auditory experience has sounds (even if some of the content of an auditory and visual experience overlap). Indeed, perhaps all of the experiential differences between visual, auditory, etc. experiences are a function of content (Speaks 2015, ch. 24–26; see also Chalmers 2004). But if it were reasonable to maintain that talk of visual, auditory, etc. experiences refers to a single experiential way of having content that is uniform across different sensory experiences, then presumably it is likewise reasonable, for all we’ve seen, for perceptualists to maintain that talk of anger, sadness, etc. refers to a single attitude that is uniform across different emotions. If this were correct, then the Perceptualism as a Bad Start Objection would fail to gain independent leverage insofar as it stacks the deck by inviting a tendentious comparison between emotions (emotional experiences) and perception, belief, desire, etc. rather than visual, auditory, tactile, etc. experiences.

But is it plausible that different sensory modalities involve a single experiential way of having content?¹⁵ One important argument for an affirmative answer builds on the phenomenon of perceptual binding.¹⁶ To illustrate, suppose a person sees a basketball as orange and spherical. They don’t just simultaneously see something orange and something spherical but rather experience a single entity as orange and spherical. This is intramodal perceptual binding. Such binding can also occur intermodally. For example, one may perceptually experience a brown dog as barking (Speaks 2015, 180). This isn’t merely the co-occurrence of a visual experience as of a brown dog at a certain location and an auditory experience as of barking nearby. The brown and the barking are experienced as having a common source. But since the sound (barking) isn’t seen and the color (brown) isn’t heard, this experience seems to be intermodal in character. Following Speaks, let’s call this intermodal experience a C-representation (2015, 183–184).

Consider now the question of whether in C-representing the dog as brown and barking one likewise C-represents the dog as brown and C-represents the dog as barking. There is pressure to say yes. To see this, consider how other attitudes work. For example, if one believes that the dog is brown and barking, then one believes that the dog is brown and believes that it is barking. Or returning to the example of intramodal binding, in seeing the basketball as orange and spherical, one sees the basketball as orange and sees it as spherical. Barring a persuasive argument to the contrary, we should likewise say that C-representations distribute over conjunction in just the same ways as believing and seeing. But now it looks like C-representations are, as Speaks puts it, “swallowing up the other species of perceptual representation” (2015, 184). Rather than insisting C-representations occur alongside visual, auditory, etc. experiences with the same content, Speaks suggests that there is a single experiential way of having content common to each.¹⁷ In other words, visual, auditory, etc. experiences aren’t each distinctive attitudes in their own right; they rather qualify a singular perceptual attitude. While I cannot fully investigate the prospects for this view here (though section 4 addresses an important objection that parallels another of the Attitudinalist Objections), it offers an attractive framework in which to develop perceptualism.

Faced with the Perceptualism as a Bad Start Objection, then, perceptualists should say that just as different sensory experiences involve the same underlying attitude, so too do emotions. Moreover, on a straightforward version of perceptualism, the experiential way of having content implicated in emotions is the same as that involved in sensory perceptual experience. Such a view pairs naturally with the standard perceptualist idea that describing what it is like to have an emotional experience requires reference to value just as describing what it is like to have a sensory experience requires reference to what the sensory experience is about. I referred to this perceptualist idea above as the inseparability of emotional phenomenology and value.¹⁸ According to the present proposal, this similarity between emotional and sensory experience is explained by the fact that the experiential way of having content is the same in each case.¹⁹ Of course, this doesn’t mean that there won’t be differences. For just as experiencing an odor is very different from experiencing a sound, so too is experiencing value very different from experiencing either. But the perceptualist position is that these are differences in content rather than differences in experiential ways of having content. In sections 4 and 5 below I’ll have more to say on how perceptualists can theorize about these ways of having content.

Before moving on, it’s worth noticing that perceptualism’s fate isn’t necessarily beholden to the view that emotions involve the same way of representing as ordinary sensory experience. Nevertheless, if a perceptualist doesn’t follow this path, it raises concerns about whether they will ultimately have an adequate response to the Perceptualism as a Bad Start Objection. Such a perceptualist has two options. On the one hand, they may say that talk of different emotions refers to a single attitude of “emoting.” But then perceptualists would face the burden of saying what such emoting consists in, including how it is distinct from the attitude implicated in ordinary sensory-perceptual experience (Deonna and Teroni 2012, 78). On the other hand, a perceptualist could observe that perceptualism is compatible with taking different types of emotions to be distinct attitudes. That is, a perceptualist may argue that fear is a fearful representation of something as dangerous, anger an angry representation of something as offensive, and so on. The difficulty here is that it isn’t clear what an angry or fearful way of having content is. Analyzing them in terms of their corresponding values may seem objectionably redundant, given that those values are already in the content. And taking them to be primitive ways of representing strikes me as theoretically disappointing, best reserved as a last resort. Taking seriously the Perceptualism as a Bad Start Objection thus pressures perceptualists to maintain that emotions and sensory experiences involve the same experiential way of representing.

4 The Second and Third Attitudinalist Objections: Portable Contents and Fading Emotions

4.1 Unpacking the Objections

I turn now to the second dimension of the Attitudinalist Challenge, which consists of two related arguments. Answering these objections reveals hitherto underappreciated points of disagreement between attitudinalism and perceptualism. This will take some work to see, however, since Deonna and Teroni’s arguments may initially appear question-begging.

To begin, Deonna and Teroni (2015, 297) observe that we often talk as if distinct emotions are about the same thing. For example, we might say that one person is angry about something that another finds amusing. But perceptualism denies this insofar as it ascribes different content to anger than it does to amusement. Put generally, the objection is as follows. The first premise is that different types of emotion can be about the same thing. The second premise is that if instances of different emotion types can be about the same thing, then emotions as such do not contribute anything to what is represented. But then this is a problem for perceptualism, since perceptualism says that each emotion is tied to a corresponding value that it represents. In other words, perceptualism is committed to the following claim that the attitudinalist rejects: the full content of one emotion type (anger) is never entirely portable to another emotion type (e.g., amusement). Call this the Portable Contents Objection.

Deonna and Teroni offer what they take to be a similar argument using an example involving a single emotion. Here is what they say:

Maurice is not amused anymore by Barbara’s excellent joke for he heard it a hundred times. This is because his attitude towards the joke has changed, not because of a change in the content of the joke. We expect Maurice to insist that the joke is very funny while stressing the fact that at that point he heard it too many times (Herzberg 2012, 81). We have no apparent reason to think that these everyday situations imply a difference in what the subject’s mind is concerned with as opposed to the way his mind is concerned with it. (Deonna and Teroni 2015, 297)

This example involving a single emotion is meant to illustrate that emotions can come and go without changing what one represents. Maurice continues to represent the joke as funny—presumably by way of a belief—even as his amusement fades. Although Deonna and Teroni group this objection with the Portable Contents Objection, it will, for reasons that become clear below, be worth keeping separate. I call this the Fading Emotions Objection.

These objections may appear question-begging. As Mauro Rossi and Christine Tappolet point out in their defense of perceptualism, we must not conflate what they call the intentional object of an emotion with its entire content (Rossi and Tappolet 2019, 552). The intentional object of, say, Maurice’s amusement at Barbara’s joke is the joke itself. But then the perceptualist adds to this a story about what amusement is, namely an experience of its object as amusing. So for Deonna and Teroni to insist that different emotions can have the same content is to beg the question. And Rossi and Tappolet could add that in cases where amusement fades (though they don’t address cases of this sort directly), we must not simply assume that nothing changes about what the agent represents. The perceptualist will say that even if the agent continues to believe that the joke is funny once the amusement has faded, they no longer emotionally experience it as such. In other words, what they once represented in two ways, namely through judgment and emotion, they subsequently only represent in one.

It turns out, however, that the Portable Contents Objection (and similarly the Fading Emotions Objection) can be further developed in a way that isn’t question-begging. One possibility, suggested by Rossi and Tappolet (2019), is that the objection may proceed from general commitments about the nature of formal objects, and the formal objects of emotions in particular (see also Deonna and Teroni 2012, 76). Formal objects are distinguished from intentional objects, or particular objects (see Kenny 1963; Teroni 2007, 396). In general, formal objects “are supposed to shed light on specific categories of mental states” (Teroni 2007, 396). For example, the intentional object of a belief that \(P\) is \(P\), but the formal object, at least according to one common view, is truth. Whereas \(P\) can figure in the content of many different mental states (e.g., one can suppose that \(P\)), the formal object, truth, seems to tell us something important about the nature of belief itself. Similarly, according to a familiar story about emotions, the formal objects of emotions are the values corresponding to each emotion. Fear of a bear, say, has two objects: the intentional object is the bear and the formal object is danger. Such formal objects perform at least two main tasks (Rossi and Tappolet 2019, 549). First, they help to determine an emotion’s correctness conditions. Fearing that \(P\) is correct just in case \(P\) is dangerous. Second, the formal object individuates the type of emotion in question. For example, anger is distinct from fear because these emotions have distinct formal objects.²⁰ Rossi and Tappolet then point out that, according to Deonna and Teroni, the formal object of an attitude is never part of its content (Deonna and Teroni 2012, 76). This picture of formal objects, then, denies the perceptualist any gap between the intentional object of emotions and the “entire” content of emotions. So when we say, for example, that one person is angry about what another person finds amusing, perceptualism can’t make sense of this. And so, the thought goes, attitudinalism is a better starting point for emotion theory.

Rossi and Tappolet offer a reply on behalf of perceptualism. Their reply begins by conceding that the formal object of many mental states resides outside those states’ content. For example, a belief that \(P\) doesn’t represent that \(P\) is true. It’s rather that truth characterizes the correctness conditions for the attitude type rather than its content (Rossi and Tappolet 2019, 555). But according to them, the formal objects of some non-emotional attitudes do feature in those attitudes’ content. Here they point to chromatic perceptual experiences. These include visual experiences of red, green, etc. Take a visual experience of an object as red. This experience has redness as part of its content. But if formal objects individuate attitudes and determine their correctness conditions, then redness is likewise the formal object. For as Rossi and Tappolet observe, “redness is that which, in conjunction with the intentional object of a perception of red, determines whether the perception is correct or not” (2019, 551). And “redness is the property that individuates the type of perception in question, namely, a perception of red” (Rossi and Tappolet 2019, 551).

As it stands, advocates of the Portable Contents and Fading Emotions Objections are unlikely to find Rossi and Tappolet’s defense of perceptualism persuasive, and reasonably enough. This is because the notion of formal objects has arguably been cheapened to the point that they are no longer revelatory of the attitude or mental state in question (cf. Teroni 2007, 396; Müller 2017, 284). To illustrate, suppose that chromatic perceptual experiences, including “reddish” visual experiences, “bluish” visual experiences, etc., mark distinctive attitudes with their own formal objects. One may worry that, if this were the case, then there are as many distinctive attitudes and formal objects as there are properties that can be perceived. This includes not only colors such as red but specific shades of red, specific shapes, motion properties, etc. And beyond perceptual experience, if groupings of similar contents are viewed as sufficient grounds for invoking distinctive attitudes and corresponding formal objects, it isn’t clear why this line of response wouldn’t generate the result that, say, chromatic beliefs also have colors as their formal objects (perhaps in addition to truth). So Deonna and Teroni can reasonably deny that adding qualifications such as “chromatic” (or “shaped,” etc.) to “perceptual experience” and “belief” marks a new attitude with its own formal object.

As we’ll see momentarily, Rossi and Tappolet’s reply gets something importantly right. Perceptualists should take the relation between emotions and values to be analogous to the relation between chromatic perceptual experiences and colors. But perceptualists need to be cautious about the language of formal objects, perhaps even setting it aside (at least initially) as something that tends to obfuscate the most natural ways of framing perceptualism. The perceptualist reply that I offer to the Portable Contents and Fading Emotions Objections emerges by attending in the right way to the core comparison between emotions and sensory experiences that motivates perceptualism in the first place.

4.2 Answering the Portable Contents and Fading Emotions Objections

Perceptualists can still answer the Portable Contents and Fading Emotions Objections, but doing so requires being careful about the contemporary dogma that the formal objects of emotions are corresponding values. A bit of extra terminology will help to clarify the dialectic. This is the language of representational guises, a notion with roots as far back as Aquinas (see Tenenbaum 2006). The intuitive idea is that a representational guise is a way of representing that “casts” content in a certain light. Here is how Kriegel describes such castings:

I propose that we capture this by saying that when a mental state represents \(p\) under the guise of the \(F\), the state does not represent \(p\) as \(F\), but rather represents-as-\(F\) \(p\). Thus, a belief that \(p\) does not represent \(p\) as true, but represents-as-true \(p\). That which it represents is simply \(p\). Representing-as-true is a way, or mode, of representing the mode characteristic of belief. (Indeed, it would not be far-fetched to hold that believing just is representing-as-true.) What this means is that in representing \(p\) under the guise of the true, the belief that \(p\) represents \(p\) in a “truth-committal” manner. It takes a truth-y stance toward \(p\). Similarly, a desire that \(p\) does not represent \(p\) as good, but represents-as-good \(p\). (Kriegel 2019b, 10) ²¹

These remarks indicate a close relationship between the role of representational guises and those often assigned to formal objects. Whereas Deonna and Teroni invoke formal objects to distinguish the attitude of belief from that of desire, Kriegel invokes representational guises to make this distinction. Indeed, Kriegel is explicit that (at least for some attitudes) he recommends conceiving of the property typically cited as the formal object as the representational guise (2019b, 16).

Perceptualists, however, should distinguish between representational guises and formal objects. For the sake of sticking with the custom in emotion theory, they can continue to treat an emotion’s formal object as its corresponding value. But then what about the representational guise of emotions? The answer is almost irresistible. After all, the view is called perceptualism. As we have seen, the natural perceptualist response to the Perceptualism as a Bad Start Objection says that emotions involve the same experiential way of having content as paradigmatic perceptual experiences. The notion of a representational guise offers a more concrete understanding of this proposal. That is, perceptualism pairs naturally with the view that emotions have the same representational guise as ordinary perceptual experience. One natural candidate for the guise involved in perceptual experience is the following: representing-as-present (cf. Kriegel 2019a, 159–160).²² The idea here is to capture an important feature of the phenomenology of perceiving, namely that in perceiving one has an impression of certain objects and properties as being present; and when a perceptual experience is veridical, one is acquainted with those very properties.²³ So, on this proposal, in perceiving the brown dog one stands in a relation to the dog such that one represents-as-present the brown dog. Similarly, fearing the dog might consist in representing-as-present the dangerous dog.

One worry about this proposal stems from the temporal orientation of some emotions.²⁴ To illustrate, it might seem as if sadness and fear can’t represent-as-present since sadness is about the past and fear about the future. But on closer inspection, there’s no immediate cause for concern here. For even if such emotions include in their cognitive bases thoughts directed to the past or future, it doesn’t follow that the evaluative properties that they represent would not be present.²⁵ Consider a person who is sad about having been fired from work. This past event can explain things, most obviously certain absences, that matter now (e.g., an absence of fulfilling work). Furthermore, and in general, when a past event ceases to explain anything of negative value in the present (e.g., one finds a better job), one is typically no longer sad, or at least it seems fitting not to be; and so it strikes me as prima facie plausible that sadness represents-as-present some negative value (typically grounded in an apparent absence explained by a past event).²⁶ A similar point works for fear, as already indicated by the brown dog example above. In particular, while fear can be driven by thoughts of a possible future outcome, it is the prospect of that outcome now that makes something dangerous. In general, then, sadness and fear aren’t obviously exceptions to the proposal that emotions represent-as-present. Of course, whether certain emotions are temporally oriented in such a way that they can’t be understood to represent-as-present value depends on a detailed study of particular emotions. And while I’m optimistic such explorations will vindicate the present proposal, this is beyond what I can hope to accomplish here.²⁷

Whatever one thinks about this specific proposal about the guise involved in perceiving, however, the big picture perceptualist idea is just this: emotions have that very same representational guise as perceptual experience. So insofar as it seems as if formal objects are revelatory of the nature of attitudes, rather than the content of attitudes, this is because we are overlooking a key point: perceptualism naturally generates a key distinction between an emotion’s representational guise and its formal object.²⁸ The former is common to all emotions while the latter is distinctive of the emotion type in question.

We’re now positioned to see how the perceptualist ought to respond to the Portable Contents and Fading Emotions Objections. Let’s start with the latter. In presenting that objection, recall that Deonna and Teroni describe Maurice’s fading emotional response to Barbara’s joke. Despite no longer being amused by the joke, he still believes that it’s funny. They say, “The fact that an evaluative property features in the content of a mental state is hardly sufficient to make it an emotion, let alone an emotion of a specific type” (Deonna and Teroni 2015, 297). But now consider an analogous argument centering on perceptual experience. In particular, take the following, clearly misguided, objection to the view that a perceptual experience as of a red car represents redness (which parodies Deonna and Teroni’s statement of the Fading Emotions Objection; cf. Deonna and Teroni (2015), 297):

But this objection doesn’t work. This is because visual experiences involve a distinctively perceptual way of representing certain contents that is importantly different from the way contents are represented in belief. On one view, the difference between perceptual and cognitive ways of representing is primitive, at least on the phenomenal level we’re concerned with here (see Kriegel (2019a)). But the notion of representational guises offers hope for (at least partially) analyzing this difference. For example, following Kriegel’s suggestion above, and in accord with those who take the formal object of belief to be truth, we may say that believing that \(P\) is a matter of representing-as-true \(P\) (Kriegel 2019b, 10; see also Deonna and Teroni 2015, 308).²⁹ By contrast, perceptual experiences are plausibly oriented to objects and properties, which are more aptly described as present rather than true.

Turn now to the Portable Contents Objection. Recall that, according to this objection, everyday discourse about emotions suggests that different emotions can be about the same thing. For example, we might say that one person is angry about what was amusing to another. But if different types of emotions are about the same thing, then, contrary to perceptualism, emotions don’t contribute anything to what is represented. To see why this objection shouldn’t persuade us, turn once again to ordinary sensory experience. We might say that while Cassandra heard the ambulance approaching, Benny saw the ambulance approaching. The presence of this common content paired with the difference in the phenomenology of the two experiences, may tempt one to conclude that vision and audition are different experiential ways of having content. But this inference would be a mistake.³⁰ The reason is because Cassandra’s auditory experience and Benny’s visual experience only have overlapping content, not the same content. After all, Cassandra’s experience included various sounds as part of its content while Benny’s included various colors and shapes. And perceptual experiences with color content have a very different phenomenology from perceptual experiences with sound content. So when we transfer the reasoning behind the Portable Contents Objection to the perceptual case, the argument fails to show that experiences in different sensory modalities can share their entire content. The Portable Contents Objection, then, really only shows that emotions have overlapping contents, and perceptualists agree with that.

The perceptualist position being proposed here can be further illustrated by way of comparison with the attitude of disbelieving.³¹ For example, one might say that Obama disbelieves what Trump believes. Here the content of the disbelief and the content of the belief are not exactly the same. This is because “disbelieves” refers to both an attitude as well as a content, perhaps among other things. In particular, it seems to be a shorthand way of referring to a belief that something is not the case (see Price 1989, 120–121). The perceptualist thinks that talk of emotions functions similarly. That is, talk of sadness, anger, joy, etc. refers both to an attitude as well as a content; and it’s the content represented under a certain guise that makes a given emotion the emotion that it is.³²

5 The Fourth Attitudinalist Objection: Standards of Correctness

If what I have argued so far is correct, then perceptualists can also answer the fourth and final Attitudinalist Objection, what I call the Standards of Correctness Objection. According to this objection, the attitudinal theory better explains the correctness conditions for emotions. By way of comparison, consider that a belief that \(P\) is correct just in case it is true that \(P\). Similarly, a desire that \(P\) is correct just in case it is desirable that \(P\) (or, alternatively, good that \(P\)). The different correctness conditions for the belief and desire are, according to many, explained by the nature of the respective attitudes rather than their contents. After all, as Deonna and Teroni point out:

[F]ew philosophers go along with Davidson in insisting that believing requires representing a proposition as true, or that desiring requires representing a proposition […] as desirable (Deonna and Teroni 2015, 298) ³³

Given a rejection of the Davidsonian approach, they then draw the connection to emotions:

This encourages the thought that a distinction between the respective contributions of content and attitude to the correctness conditions akin to the one sketched above for belief and desire also holds true for the emotions. To the question: “Why is fear or anger correct if the object or situation to which these emotions are directed is dangerous or offensive?”, the straightforward answer is “Because one has the attitude of fear or anger towards it” and not “Because it is represented as being dangerous or offensive.” (Deonna and Teroni 2015, 299)

The first point to notice is that Deonna and Teroni seem mistaken in an assumption about perceptualism. They take it as a data point that fear is a correct response to what is dangerous for the trivial reason that one has the attitude of fear toward it (Deonna and Teroni 2015, 299). They also suggest that perceptualists are barred from saying as much. But perceptualists can say this. Of course, they also happen to think that what fear consists in is a perceptual way of representing its object as dangerous, in a manner similar to how a visual experience of redness involves a perceptual way of representing its object as red. It is this feature of fear that helps us to understand more deeply why fear is a correct response to what is dangerous.

The core of Deonna and Teroni’s objection, however, is that attitudinalism does a better job of respecting the contributions of both attitude and content to the correctness of an emotion. For example, a belief that \(P\) is correct just in case it is true that \(P\). The content identifies a certain proposition while the attitude (belief) requires that the proposition be true. Similarly, a desire might be thought correct just in case its content is good; and so on for other attitudes. If this is how it works for other attitudes, shouldn’t it be the same for emotions? Fearing that \(P\) is correct if and only if \(P\) is the case and \(P\) is dangerous; anger that \(P\) is correct if and only if \(P\) is the case and \(P\) is offensive; and so on for other emotions.

But if what I argued in the previous section is on track, then perceptualists needn’t deny that attitude and content both contribute to the correctness conditions of emotions. Perceptualists should say that emotions share their representational guise with ordinary sensory experiences, and this guise contributes to the correctness conditions of different emotions. This is not to my knowledge a point that perceptualists have emphasized.³⁴ But it’s hard to overstate how natural it is for a perceptualist about emotions to say this in response to the Standards of Correctness Objection. Incidentally, this is also what perceptualists about desire should say. That is, philosophers who maintain that desires are a perceptual representation of some normative property or relation can say that desires represent-as-present their contents (e.g., Oddie 2005). This proposal on behalf of perceptualism about desire, as with emotion, concerns the total content of the desire. Perceptualists about emotion/desire think that talk of emotion/desire refers both to an attitude and its proprietary content, each of which make contributions to the correctness conditions of the attitude. And as we saw in the last section, there is nothing obviously ad hoc about taking talk of emotions, or desires for that matter, to refer both to attitudes and contents.

Over the course of the last two sections, I have argued that perceptualists should draw a perhaps surprising distinction between an emotion’s representational guise—treating it as identical to perceptual experience—and its formal object—taking it to be a value proprietary to the type of emotion in question. I close this section by raising a question about whether attitudinalists may have reason to adopt their own distinction between representational guises and formal objects.³⁵ Whether they do may depend on whether they agree with perceptualists about a key dimension of how perceptualists characterize emotional phenomenology. On the view sketched here, perceptualists maintain that emotions share a representational guise with sensory experience, namely that certain content is represented-as-present. This is a way of unpacking Döring’s thought that emotions and perceptions put forth certain contents as actually there (Döring 2007, 377). Emotional experience is thus unlike (voluntary) imaginative experiences, or suppositions, which do not put forth their contents in this way, and therefore imagination and supposition do not have correctness conditions mirroring that of perceptual experience and emotion. A question thus arises for attitudinalists about whether they would agree with those perceptualists who take emotions to put forth their contents as present. And if so, then there is reason for the attitudinalists to complexify what they take the formal objects of emotions to be, or alternatively to draw their own distinction between representational guises and formal objects.³⁶ The aim here is not to present an objection to attitudinalism but rather to raise a question that helps us to better frame the possible points of (dis)agreement between perceptualism and various versions of attitudinalism.

6 The Choice between Perceptualism and Attitudinalism

This paper has taken for granted the popular position that emotions are evaluative experiences. The aim has been to show that the interlocking objections comprising the Attitudinalist Challenge do not establish attitudinalism as a better starting point for this position. In this final section, I explain why we might ultimately favor perceptualism over Deonna and Teroni’s version of attitudinalism (i.e. action tendency attitudinalism).

As we’ve seen, action tendency attitudinalists maintain that emotional attitudes consist in feelings of readiness to act, and these feelings explain why emotions count as evaluative experiences. Here is how Deonna and Teroni describe their position:

Fear of a dog is an experience of the dog as dangerous insofar as it is an experience of one’s body being prepared to forestall its impact (flight, preventive attack, immobility, etc.), an attitude it is correct to have if, and only if, the dog is dangerous. (Deonna and Teroni 2015, 303; see also Deonna and Teroni 2012, 81)

Deonna and Teroni also maintain that there is a non-contingent connection between the experiential dimension of an emotion and its correctness conditions:

The body is felt in the form of a gestalt of bodily sensations, which consists in being ready to respond in a given way to the object. If experiencing such an attitude is all there is to experiencing something in evaluative terms, then of course the relation between the attitude and the fact that the evaluative property enters into the correctness conditions of the mental state is anything but contingent. (Deonna and Teroni 2012, 87)

The bodily sensations in fear, for example, are such that they necessarily count as experiences of their object as dangerous; and this is why fear has the correctness conditions that it does. To motivate this thought, they point out that it isn’t intelligible that amusement could be a way of making danger manifest. Given the nature of fear, it seems as if that is the only emotion that could be an experience of danger (Deonna and Teroni 2012, 86).

A major challenge for action tendency attitudinalism is to demystify how emotional experiences count as evaluative experiences. Such evaluative experiences aren’t simply a matter of covariation:

[T]he connection between the emotional experience and the evaluative property cannot be modeled on that between smoke and fire, namely as one of natural co-variation. Experiencing the evaluative property of an object is not taking the way one’s body feels as an indication, a sign, or a symptom of the fact that this object has this property. (Deonna and Teroni 2012, 87).

Deonna and Teroni argue that a covariational conception of the link between emotion and value fails to capture the thought that emotional experiences involve a presentation or manifestation of value.³⁷ My concern, however, is that action tendency attitudinalism may ultimately turn out to be, in an important sense, a version of the covariation model (perhaps a kind of necessary covariation), ultimately failing to deliver anything like a presentation of value.

Notice first what the action tendency attitudinalist isn’t saying. First, and most obviously, they aren’t saying what perceptualists say. A perceptualist, as we’ve seen, says that danger features in experiences of fear similar to how empirical properties feature in sensory experience (e.g., Roberts (2013, 72–73); Tappolet (2016, 26–28); inter alia). Such a view thus well-suited to make sense of the idea that values are manifest in emotional experiences. But Deonna and Teroni deny that emotions make value manifest in this way (2012, 68–69).

There is another important view in the vicinity of perceptualism that likewise isn’t the action tendency attitudinalist’s. This view can be understood as adapting the proposal sketched above about the representational guise of perceptual experience. According to that proposal, a full description of a perceptual experience requires reference to an attitudinal phenomenology of representing-as-present (a being-present-y mode of representation; cf. Kriegel (2019b, 10)). Building on this thought, an attitudinalist might then take emotions to have evaluative representational guises in the manner that perceptual experiences have a representing-as-present guise. Fear, for instance, might be thought to have an attitudinal phenomenology that must be described as representing-as-dangerous.³⁸ But action tendency attitudinalists don’t have in mind representational guises of this sort, either (see Kriegel 2019b, 13).³⁹ Instead, the action tendency attitudinalist maintains that the phenomenology of emotional attitudes is properly described in terms of one’s body being activated in a particular way rather than in evaluative terms. Deonna and Teroni point to the following passage from Nico Frijda to unpack their view:

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. (Frijda 2005, 382; quoted in Deonna and Teroni 2015, 308, n19)

Contrast this with the view of perceptual experience offered in section 4: whereas attending to a perceptual experience, according to that proposal, involves attending to the property of being present as a dimension of attitudinal phenomenology, the action tendency attitudinalist doesn’t think that attending to emotional experience involves attending to value as a dimension of attitudinal phenomenology.

So how exactly does the action tendency attitudinalist understand emotions as evaluative experiences? As we’ve seen, Deonna and Teroni say that emotions are “a gestalt of bodily sensations, which consists in being ready to respond in a given way to the object” (2012, 87). For example, a person who fears a snarling dog may have an experience of their body shrinking away from the snarling dog. But it’s not clear that this makes sense of emotions as evaluative experiences, or as manifesting value. Even if we add that the action tendencies associated with different emotions are (necessarily) correct responses to the relevant value, it wouldn’t thereby follow that emotions are evaluative experiences. But consider the following: might it be that emotional experiences are evaluative but don’t seem evaluative when we attend to them?⁴⁰ We can see the difficulty with this proposal by returning to Frijda’s example of pointing quoted above (2005, 382). Following Frijda, Deonna and Teroni appear to think that in attending to what it feels like to point, the experience seems to just be that of one’s finger being stretched. But notice that attending to the entirety of the experience isn’t describable simply in terms of the experience of a stretching finger. And even if we attend to the experience in abstraction from what is being pointed to, we aren’t left with merely an experience of a stretching finger. This is because a crucial part of the experience of pointing is an experience of indicating, and we can attend to this dimension—either in isolation or in conjunction with an object. So if the pointing case provides a model for emotions, then, contrary to what Deonna and Teroni suggest, a description of what we’re attending to in emotional experience—even in isolation from the emotion’s object—should require reference to an experience of value. But if the action tendency attitudinalist says this, then they have drifted in the direction of the sort of perceptualist-adjacent phenomenology they want to resist, namely one that retains a representational mode phenomenology even in higher-order attention on the experience itself. So unless the action tendency attitudinalist can somehow make sense of emotional experiences as evaluative experiences that don’t seem evaluative when we attend to them, there is pressure to give up the view that emotions are evaluative experiences.

But how much does it matter whether action tendency attitudinalism can make sense of emotions as evaluative experiences? The answer depends on what one hopes to accomplish with a theory of emotions. For example, one may be tempted by the view that evaluative knowledge is ultimately rooted in evaluative experiences. Or, more modestly, one may think that evaluative experiences are an important route to evaluative knowledge. And mental states like emotions provide a tempting non-mysterious source for what such value experiences might be (Roberts (2013); Tappolet (2016); Milona (2016); inter alia).⁴¹ Furthermore, perceptualists are often attracted to the idea that emotions are able to rationalize action and maintain, moreover, that perceptualism can explain how this is possible. We might appeal to fear, for instance, to explain a person’s fleeing a bear. If fear is an experience of its object as dangerous, then this renders the action intelligible (Döring 2007). Yet, again, if emotions aren’t evaluative experiences, if they are mere felt tendencies to act, then it is not clear that they can rationalize action (as opposed to merely cause it).

7 Conclusion

This paper has explored the Attitudinalist Challenge to perceptualism. The objections comprising the challenge are meant to illustrate that much of our pretheoretical discourse about emotions conflicts with the perceptualist theses that emotions have, and are individuated by, evaluative content. However, the Attitudinalist Challenge is unpersuasive. Still, adequately addressing the objections requires perceptualists to present their view with greater clarity. In particular, the version of perceptualism presented here draws a crucial and perhaps surprising distinction between an emotion’s representational guise, which is uniform across emotions and other perceptual experiences, and its formal object, which is specific to that emotion type. This version of perceptualism emerged in large part by comparing emotions and sensory perceptual experiences, and to this extent marks a natural development of the theory.

Acknowledgements

For extremely valuable feedback on this paper, I am grateful to Keren Gorodeisky, Hichem Naar, Katie Stockdale, Christine Tappolet, Mark Thomson, audience members at the Thumos Seminar (University of Geneva), and three anonymous referees.