The Receptive Theory: A New Theory of Emotions

Abstract

Cognitive Theories of emotions have enjoyed great popularity in recent times. Allegedly, the so-called Perceptual Theory constitutes the most attractive version of this approach. However, the Perceptual Theory has come under increasing pressure. There are at least two ways to deal with the barrage of objections, which have been mounted against the Perceptual Theory. One is to argue that the objections work only if one assumes an overly narrow conception of what perception consists in. On a better and more liberal understanding of perception, the objections lose their force. The other is to stress that the differences between emotions and sensory perceptions can be explained by focusing on a new analogy. As I will argue, emotions have interesting similarities with magnitude representations, such as the representation of distance. Such representations are plausibly thought to be analog and non-conceptual, but by contrast to sensory perceptions, such as colour perceptions, they do not lie at the sensory periphery. This new analogy makes room for a novel and attractive theory of emotions, the Receptive Theory, which allows for a positive and epistemologically fruitful characterization of emotions.

1. Introduction

The analogies between emotions and sensory perceptions have attracted the attention of many emotion theorists (e.g., [1,2,3]. In particular, the thought that emotions allow us, in some way to be explained, to ‘see’ values has a long history, which goes back to the idea of a moral sense discussed by Scottish moralists, i.e., Francis Hutcheson, David Hume, Adam Smith, and Thomas Reid. The thought that emotions play a central role in our epistemic access to the evaluative, or more broadly to the normative, has been deemed central to an account of emotions that has gained popularity over the last two decades, i.e., the so-called Perceptual Theories. The central claim of Perceptual Theories is that emotions have to be modelled on perceptual experiences. Perceptual Theories can be considered to constitute the most attractive versions of Cognitive Theories, an approach that can be traced back to the Stoics, and which became mainstream in the late 1950s, when it became clear that emotions are not mere feelings. Even among its detractors, the Perceptual Theory has attained the status of counting as one of the standard accounts of emotions (see inter alia [4,5,6,7]).

The master argument in favour of the Perceptual Theory is simply that emotions and sensory perceptual experiences share important similarities. To get a sense of that argument, think of your experience of awe when looking at a clear sky and compare it to the visual experience of the colour of the same sky. Impressed by this analogy, Alexius Meinong writes: “[w]hen I say, ‘The sky is blue’, and then say, ‘The sky is beautiful’, a property is attributed to the sky in either case. In the second case a feeling participates in the apprehension of the property, as, in the first case, an idea does. And it is natural to let the feeling be the presentative factor in the second case […]” ([2] (p. 28) my italics). The similarities between emotions and perception engender an argument from analogy that suggest that emotions have to be modelled on perception. However, a dilemma arises as soon as you realize that there are also many differences between emotions and sensory perceptual experiences. Should we opt for the Perceptual Theory in light of the important similarities or should we give it up on account of the many differences?

This dilemma nags perceptual theorists, but, in fact, all emotion theorists should worry. After all, they all have to make room for the undeniable similarities as well as the numerous differences between emotions and perceptions. There are at least two strategies in response to this dilemma. The first one consists in both minimizing the importance of the differences between emotions and perceptions and advocating a liberal conception of perception [8]. Instead of sticking to your guns, you can opt for a second strategy, which consists in developing an account of emotions that aims at holding onto the attraction of the Perceptual Theory while making room for the differences between emotions and perceptions [9,10]. I want to explore this more concessive strategy here.

This paper is structured as follows. Section 2 presents the Perceptual Theory and its main versions and discusses the argument from analogy that has been presented to defend it. The intermediate conclusion is that it would be nice to have a different theory of emotions, the price of the Perceptual Theory being possibly quite high: it involves assuming an account of perception that might seem much too lax. Section 3 introduces a new account of emotions, the Receptive Theory, according to which emotions have to be modelled on analog magnitude representations. Finally, Section 4 considers the question of whether you can nonetheless consider emotions to play an epistemic role in the justification of evaluative beliefs if one opts for the Receptive Theory. The conclusion wraps up the discussion by underscoring several questions that need further exploration.

2. The Perceptual Theory and its Many Woes

Emotions make for an intractable terrain. One difficulty arises from the fact that there are several types of affective states. It has become common to distinguish between emotions, sentiments, and moods, which each call for a different account [4,9]. Another distinction that matters is that between occurrent emotions, often referred to as ‘emotional episodes’, and dispositions to experience occurrent emotions [4]. Even if you stick to the former, a further difficulty arises given the huge variety of emotion kinds, ranging from fear, disgust, anger, surprise, joy, and sadness1 to admiration and awe, as well as indignation, pride, resentment, and gratitude, not to mention non-western emotions, such as Japanese amae, a pleasurable feeling of dependency [12], and Russian toska, which roughly consists in melancholy-cum-yearning, but can also stand for anguish, depression, or boredom [13]. Yet another difficulty comes from the fact that a typical emotional episode involves a plurality of components. You need to see the bear (or hear or maybe smell the bear, or else be told that there is a bear, etc.) before having any emotional reaction, and when you see it as it is facing you, you feel your heart as it starts pounding, you have a sense of imminent danger, something which comes with an avalanche of thoughts—is there a tree close by, should you curl up on the ground and play dead or should you make yourself as tall as you can and face the bear, etc.?—as well as with an imperative desire to be safe from the bear.

Almost everyone in the debate agrees that typical emotional episodes—for short ‘emotions’—involve five characteristics, which can be expressed as follows:

(Phenomenology) Emotions typically have phenomenal properties, in the sense that there is something it is like to undergo an emotion, whether this feeling is merely induced by bodily changes or not.

(Motivation) Emotions typically involve motivation, in the sense that they come with specific goals, e.g., safety, and correlated desires, e.g., running away from danger.

(Intentionality) Emotions typically have intentional objects, such as the bear in the case of that fear you felt on the hike.

(Evaluation) Emotions typically involve appraisals, in the sense that in some way or another, you assess the intentional object of your emotion negatively (fearsome, offensive, disgusting, etc.) or positively (admirable, prideworthy, amusing, etc.).

(Formal objects) Emotions typically have so-called ‘formal objects’, that is, features in terms of which emotions are assessed as correct or incorrect, such as fearsomeness in the case of fear, disgustingness in the case of disgust.

Apart from hybrid theories, according to which emotions have several essential components, such as evaluative judgements plus motivations and feelings, a view often attributed to Aristotle [14], or beliefs and desires (e.g., Searle [15]), the main families of theories are Feeling Theories (e.g., James [16]), Motivational Theories (e.g., Frijda [17], Deonna and Teroni [4], Scarantino [18]) and Evaluative Theories (also known as ‘Cognitive Theories’), which focus on the evaluations involved in emotions. Thus, emotions have been thought to be evaluative judgements (Nussbaum [19]), evaluative thoughts (Greenspan [20]), or evaluative construal (Roberts [21]). Yet a further possibility is to opt for Perceptual Theories and hold that emotions are, in essence, evaluative perceptions.

Perceptual Theories comes in different versions, depending on whether emotions are thought to literally be perceptions or merely quasi-perceptions.2 The most popular versions are non-literal theories, according to which emotions are kinds of quasi-perceptions of evaluative facts (e.g., de Sousa [3], Goldie [23], Helm [24]). By contrast, the Perceptual Theory I defend holds that emotions are, quite literally, perceptions, or more precisely, since we want to make room for mis-firings, perceptual experiences of evaluative facts (Tappolet [8,25,26]). On this account, your anger at your friend, François, who makes fun of your fear of bears, consists in the non-sensory, affective, perceptual experience of François as offensive.

For the sake of simplicity, I will focus on the literal version, which I will simply refer to as ‘the Perceptual Theory’. The Perceptual Theory holds that emotions are affective perceptual experiences of evaluative facts involving what I shall call ‘affective properties’, i.e., properties such as being fearsome, being disgusting, being admirable, etc., a subclass of evaluative properties that wear their tight relation to affective responses on their sleeves.3

The Perceptual Theory can be summarized as follows:

(The Perceptual Theory) Emotions are perceptual experiences that non-conceptually represent their intentional objects as having affective features, such as being fearsome, disgusting, or admirable.4

The Perceptual Theory is well-placed to account for the five characteristics listed above (see [9]). However, the master argument for this account draws on the similarities between emotions and sensory perceptual experiences. What one takes to be true of sensory perceptual experiences, which I will simply call ‘perceptions’, is controversial and for the sake of argument, I shall assume a pretty standard account of what they consist in.5 Here is a list of the main similarities that engender the argument from analogy (see [8,9,25]):

• (Intentional object) Both emotions and perceptions are about things, such as bears, skies, etc.
• (Representational content) Both emotions and perceptions represent their intentional objects as being in certain ways, as dangerous, offensive, etc. in the case of emotions, or blue, round, etc. in the case of perceptions.
• (Correctness conditions) Both emotions and perceptions can be considered correct and incorrect depending on how they represent their intentional objects.
• (Phenomenal properties) Both emotions and perceptions are (typically) characterised by a way it is like to experience them.
• (Automaticity) Both emotions and perceptions arise automatically, mandatorily, or to revert to a popular expression, passively; hence, neither emotions nor perceptions are directly subject to the will.
• (World-guidedness) Both emotions and perceptions are states that are, in general, triggered by things in the world, such as bears approaching, so that both count as world-guided.
• (Recalcitrance) Both emotions and perceptions give rise to conflict with judgements, in the sense that we can feel fear while judging that there is nothing to fear just as much as we can see lines to be of the same length while judging that they are of different length, as in the case of the Müller-Lyer illusion.
• (Inferential isolation) Both emotions and perceptions contrast with judgments in that they do not form inferential networks. Depending on what you judge, you are committed, on pain of irrationality, to a number of further judgments. By contrast, you can well experience fear without being rationally required to undergo other mental states, be they judgments, perceptions, or emotions. Similarly, you can well see a red flower without being rationally required to have other visual or more generally mental states.
• (Modularity) Both emotions and perceptions bear the hall-mark of modularity: informational encapsulation, specificity, in the sense that responses are restricted to a specific class of stimuli; mandatoriness (or automaticity); opacity, that is, central nervous systems processes have no access to the information that is treated; rapidity in processing the information; superficial outputs, i.e., outputs framed in basic categories; and specificity in breakdown patterns [22,28].6
• (Non-conceptual contents) Even if some emotions have conceptually articulated contents as cognitive bases, emotions and perceptions have non-conceptual contents.7 Indeed, there is reason to hold that emotions not only have non-conceptual contents and are non-conceptual states that we can experience without possessing the relevant concepts, but also that the format of the representations is analog (see [8,25,31]).8

This impressive list easily lends itself to an argument from analogy: since emotions and perceptions share many features that appear central to their nature, one can, it seems, conclude that emotions are perceptions. Now, arguments from analogy get bad press, and rightly so. A first problem is that arguments from analogy suggest a conclusion, but they fail to establish it. Two things could share most of their features while nonetheless being different kinds of things. A second problem is that an argument from analogy leads to a conclusion that something is of the same type as something else. An emotion, we are told, is a kind of perception. But this falls short of being a detailed and distinctive account of the subject matter. We need to go beyond a list of shared features to fully characterized emotions.9

These two problems appear negligible compared to the next one, which is that when we look more closely, there are more differences between emotions and perceptions than there are similarities.10 Here is a list of the main differences that have been underscored in the literature:

(1)

(No Sensory Organ(s)) There is no question that emotions fail to be associated with sensory organs, whether it is being thought as one big emotional organ or else as a family of organs corresponding to kinds of emotions.

(2)

(Cognitive Bases) Emotions, but not sensory perceptions, require cognitive bases. As we noted, you need to see (or hear, etc.) the bear in order to experience fear.

(3)

(No Causal Constraint) While perceptions are answerable to a causal constraint in that what is perceived has to be causally responsible for the perceptions, emotions can well be about things that have no causal power, such as past events, abstract entities, and fictions.

(4)

(‘Perception’ of Normative Fact) If emotions were perceptions, we would have to conclude that they are perception of entities that not only are ontologically dubious, but also quite distinct from the stuff we can grasp with our senses, namely, normative facts, where normative is understood widely to include the evaluative. Not so with perceptions, the content of which is perfectly natural.

(5)

(No Phenomenal Transparency) By contrast to perceptions, what it feels like to experience an emotion is not best described in terms of the perceived object and its properties, as if one were to look through a window unto the world.

(6)

(Valence) By contrast to perceptions, emotions can be positive or negative. Think about the contrast between joy, admiration, gratefulness, and hope, on the one hand, and sadness, spite, fear, anger, and fear, on the other.

(7)

(Close Tie to Motivation) Most, if not all, emotions appear closely connected to motivation and action, and this makes for a further contrast between emotions and perceptions, since the latter appear to have no direct connection to motivation and action.

(8)

(Irrationality of Recalcitrance) Another alleged difference between emotions and perceptions is that by contrast to cases of conflict between a sensory perception and a judgment, emotions that conflict with evaluative judgments are deemed irrational.

(9)

(Openness to Why Questions) Emotions can be questioned in terms of their justification. You can ask your friend François why he feels fear, for instance. Perceptions, by contrast, do not seem open to such why-questions. It is strange to ask your friend why he sees the bear as brown, for instance.

(10)

(Psychological Influence) Emotions, but not perceptions, are importantly influenced by an individual’s mental states, such as their desires or expectations. If you have an intense desire to see a bear close up, you are less likely to feel overwhelming fear at the sight of that big animal on the trail.

(11)

(Socio-Cultural Shaping) A feature, which has fascinated anthropologists, is that many aspects of emotions are shaped by social-cultural learning and education. This influence concerns the stimuli that trigger a particular emotion, but also facial expression and motivation.

(12)

(Emotional Plasticity) Occurrent emotions allow for socio-cultural shaping, but more importantly, emotional dispositions are heavily influenced by socio-cultural learning. Accordingly, what we are disposed to fear or to be angry with depends at least in part on the context in which we have been socialized. Not so for sensory perceptions, at least if we rule out conceptually enriched contents.

This long list of alleged difference gives food for thought. As mentioned earlier, there are at least two strategies in response to the observation that emotions and perceptions differ in a great many respects. I used to be convinced by the first, two-pronged strategy [8]. It consists, firstly, in minimizing the importance of these differences while advocating a liberal conception of what counts as a perception. I argued, for instance, that some of the alleged differences between emotions and perception are less deep than it might seem. Thus, it is not obvious that the phenomenology of sensory experiences is as simple as the No Phenomenal Transparency thesis suggests. What we see appears influenced not only by the objects and their properties, as we discover when we remove tinted glasses. Furthermore, I argued that some differences do not seem to matter. The fact that there is or are no emotional organ(s) only entails that emotions differ from sensory perceptions, not that they fail to consist in perceptions. Similarly, the fact that emotions are valenced could well be a feature of affective perceptions that sensory perception lack.

Even if one grants that the differences between emotions and sensory perceptions are not as important as it might first seem, it is hard to deny that some real and important differences remain. Hence, the second part of the two-pronged strategy, i.e., the adoption of a liberal conception of what counts as a perception. There are, I believe, good reasons to adopt such a liberal account, for it makes room for cases such as proprioception and multi-modal perception. As such, the concept of perception is not necessarily tied to a particularly narrow conception.

More could be said in defense of this two-pronged strategy [8]. However, I came to realize that there is no need to opt for the two-pronged strategy. There is another way out, which is just as good, and even a bit better if you have worries about being a liberal about perceptions—maybe such a liberal account entails that we can literally see numbers, ghosts, and spirits, and maybe this is an undesirable implication, which would threaten healthy empiricism. This second strategy consists in developing a new account of emotions, the Receptive Theory, to which we now turn.

3. The Receptive Theory11

The Receptive Theory12 is faithful to the gist of the Perceptual Theory, for it holds that emotions are non-conceptual representations of evaluative, or more precisely affective, facts. But it concedes that emotions are not perceptions, or at least that they are not perceptions in any strict sense. The receptive theory is thus premised on the acceptance of a strict conception of perception, according to which perceptions have the two following features:

(a)

(Stimulus Dependence) Perceptions are stimulus-dependent, which is to say that they satisfy the aforementioned causal constraint;

(b)

(Sensory Modality) They are tied to a sensory modality, which is to say that they are connected to a sensory organ.

As we have seen, emotions have cognitive bases, so they fail the stimulus dependence requirement. Furthermore, there is no sensory modality that is tied to emotions. Accordingly, the Receptive Theory holds that emotions are not, in any strict sense, perceptual experiences.

The good news is that acknowledging these two points is perfectly compatible with holding on to the claim that emotions have representational contents. Thus, the Receptive Theory claims that emotions are non-conceptual representations, in the sense that emotions non-conceptually represent their objects as having affective properties.

In order to see how this is possible, it will prove helpful to develop a new analogy. Instead of considering perception, let us look at analog representations of magnitudes, such as number, duration, distance, etc. As we will see, such representations have non-conceptual representational content but are stimulus-independent and amodal, in that they are not tied to a sensory modality. The lesson is that non-conceptual representations need not be perceptions in any strict sense. Moreover, as I will argue, the argument for the claim that such representations have non-conceptual contents appears to be transposable to the case of emotions. Here is what Jacob Beck, whose work I will extensively use here, writes:

Given their stimulus-independence and amodality, analogue magnitudes states are removed from the sensory periphery of perception. […] [N]onceptual content seeps beyond perception and infiltrates cognition.

[39] (p. 588)

So, what are analog representations of magnitudes and why should we think that they have non-conceptual contents? Consider how pigeons see the world. One striking fact is that pigeons are quite good at counting. But they do it a bit differently from what we are taught at school. They do it analogically.

Analog representations can be defined as representations that, by contrast with digital representations, involve elements that co-vary with what is represented. An analog representation involves a magnitude, such as the rate of rotation or of neural firing, which is isomorphic to what is represented. For example, a watch with rotating hands displays an analog representation of time, by contrast to the digital representation of time of the numerical display on your cell phone. As time marches on, the hand on your watch steadily rotates, co-varying with the movement of time.

Analogicity is a syntactic, not a semantic, property. Even so, there is reason to think that the syntactic property of analogicity matches the semantic property of non-conceptuality. Put differently, there is reason to think that analog representations have non-conceptual contents and vice versa, that representations that have non-conceptual contents are analog. This is so because the best explanation of why some mental states fail to satisfy a well-known semantic principle, the Generality Constraint, and thus can be thought to have non-conceptual contents, is precisely that they consist in analog representations (see [32], as well as [39]).

The Generality Constraint is a requirement cognitive states have to satisfy in order to count as having conceptually articulated content. It consists in the requirement that such a state be Systematic, i.e., that its content constituents can be recombined in any semantically acceptable way so as to form the content of other cognitive states ([40], which draws on [41]). The content of states that satisfy the Generality Constraint contains concepts, understood as ‘inferentially relevant constituents’, to use Tim Crane’s expression [40]. Suppose you believe that a crow is black. The content of this belief contains crow and black as inferentially relevant constituents. Given that the content of that belief has these constituents, it follows that if you are able to believe that the crow is black as well as to believe that the cat is ginger, you are able to form the belief that the crow is ginger, or else that the cat is black. And if you believe that the crow is black and you believe that the cat is black, you are able to infer that both the crow and the cat are black.

What is the argument for the claim that representations of magnitudes are characterized by non-conceptual contents? Beck [39] argues that such representations violate Systematicity, thus failing the Generality Constraint. The fact is that pigeons, just as any other animals (including human beings when they estimate the amount instead of explicitly measuring it), have an error rate in their grasp of magnitudes that is proportional to the size of the magnitude in question. To use the standard term in ethology, pigeons obey Weber’s law when they discriminate between magnitudes. A positive Weber constant appears to be a universal feature of magnitude discrimination by both non-human animals and human beings when they estimate magnitude. Pigeons, for instance, are able to reliably discriminate numbers only if their ratio does not exceed 9:10. Thus, experiments show that pigeons are able to represent that 38 pecks are fewer than 47, and that 40 pecks are fewer than 50, but they fail to discriminate between 38 and 40 pecks. Thus, there is reason to conclude that they are not able to represent that 38 pecks are fewer than 40. As Beck argues, this failure of Systematicity is incompatible with an interpretation of the pigeons’ numerical representations as involving conceptual contents. On the contrary, it is best explained in terms of representation that have an analog format plus non-conceptual contents.

The interesting point for us is that Beck’s argument can be transposed to the case of emotions. What needs emphasizing is that affective properties, and more generally evaluative properties, are magnitudes that form continua. Consider admirability. Clearly, something can be more or less admirable, each item that is admirable figuring on a continuum that goes from not admirable to admirable.13 Furthermore, it has long been accepted that there is a correlation between the intensity of emotions and the degrees of the relevant affective properties.14 Thus, emotions have been thought to have correctness conditions that involve not only kinds of affective properties, but also the degrees of that property.15 Such consideration of size, to use the expression coined by D’Arms and Jacobson [43], lends itself to an account of emotions as representing not only a kind of affective property but also the degree of that property. For instance, when you admire pianist Martha Argerich very much, what your admiration represents is not only that she is admirable (at least qua pianist), but also that she is admirable to a high degree. If your admiration for Mitsuko Uchida is somewhat less intense, you represent her as somewhat less admirable (at least qua pianist). Thus, each occurrence of an emotion of admiration you feel places what is admired on an admirability continuum.

Given this, it is plausible that affective representations of affective properties violate Systematicity. This can be shown using a toy example. Consider disgust felt at a platter of cheese. It might well be the case that the disgust you feel towards each piece of cheese differ in intensity. Let us suppose that there are five pieces of cheese: Appenzeller, Brie, Camembert, Dorset, and overripe Époisse, which we will call, ‘A’, ‘B’, ‘C’, ‘D’, ‘E’, respectively. As it is to be expected, your disgust directed at the overripe Époisse is the strongest. You would not have a bite of that cheese even if you were offered a fortune. On a scale of 100, you would place it just at the top, at 100. Next on the platter, the Appenzeller and the Brie have sad and soggy looks and you feel some disgust towards each of them. As to the Camembert and the Dorset, they each trigger an even more significant amount of disgust.

Given this, let us suppose that you have an error rate in your disgust reactions that is characterised by the Weber constant of 9:10 that hampered the pigeons’ ability to discriminate numbers. Thus, it could be the case that what we can call your ‘disgust module’, i.e., your ability to feel disgust and discriminate and choose between things accordingly, can discriminate and represent that (where the number represent the place on the disgustingness continuum):

(1)

A (38) is less disgusting than C (47),

and that

(2)

B (40) is less disgusting than D (50),

while neither being able to discriminate nor represent that

(3)

A (38) is less disgusting than B (40).

The possibility of this kind of discrimination failure is plausible because it is not easy to distinguish small differences in intensity when we compare emotional reactions. If Beck is right, the best explanation of such discrimination failures, which results in a violation of Systematicity, is that the representational modules at stake involve non-conceptual contents. The moral of the story, it appears, is that when we use our emotional modules to find our way in an evaluative landscape, we are more like pigeons that reasoning human beings.

There is no question that the present argument is purely speculative. By contrast to the case of pigeons’ counting abilities, there is no empirical confirmation of a Weber Constant in our uptake of disgustingness. Indeed, there are apparently important barriers to conduct an empirical validation of the hypothesis. One problem is that as my cheese platter example and the fictitious disgustingness scale vividly illustrates, there appears to be no objective scale of disgustingness. A good question here is what it would take to have such a scale, and more pointedly, whether there could indeed be a scale that lends itself to empirical measures. If disgustingness is an entirely relative property, something not being disgusting per se but only relative to an individual, we could at most measure individual disgustingness scales. The case would be comparable to that of preference scales, which are indexed to individuals. Because preferences can be studied empirically, I see no principled obstacle to conduct an empirical study that could confirm the hypothesis that emotions are analog representations having non-conceptual contents.16

But even in the absence of an empirical confirmation of these speculations, it can be agreed that the claim that emotions are informational states that have analog format and nonconceptual contents is plausible. Given this, there is reason to think that the Receptive Theory is correct. Let me spell out central claims of the Receptive Theory:

(The Receptive Theory) Emotions are representational states that have

(a)

correctness conditions involving affective properties (i.e., admirability, disgustingness, fearsomeness, etc.);

(b)

analog format and non-conceptual representational content;

(c)

cognitive bases, thus being stimulus-independent and amodal.17

In a nutshell, emotions are analog and non-conceptual, but not perceptual, representations of things as having specific affective properties. The reason why they do not count as perceptions is that they are located at one remove from the sensory periphery, given their dependence on cognitive bases.

Let me close this section with a brief list of the main attractions of the Receptive Theory. The first attraction is that the new theory holds on to the main virtues of the Perceptual Theory. One point to underline is that according to both the Perceptual Theory and the Receptive Theory, emotions are representations that can be assessed in terms of correctness. Another point is that the Receptive Theory has no more trouble than the Perceptual Theory to explain the difference between emotions and evaluative judgments in terms of the analog format plus non-conceptual content. Similarly, because emotions concern evaluative features, the Receptive Theory can recycle the Perceptual Theory’s explanation of how emotions can play a motivational role. Insofar as emotions have the function of informing us about evaluative features, it is easy to see that at least most of them tend to have an impact on what we are inclined to do.

In addition to this, however, the Receptive Theory is also well-placed to account for the differences with perceptual experiences. This should not come as a surprise. After all, the Receptive Theory is tailored to account for features of emotions such as stimulus independence and amodality. Because emotions are thought to be at one remove from the sensory periphery, the account has no worries regarding the absence of corresponding organs, the dependence on cognitive bases as well as the fact that emotions need not be causally correlated with their objects. Similarly, the emphasis on the cognitive bases of emotions allows the theory to make room for the fact that emotions are shaped by cultural and social factors, the association between sets of cognitive bases and emotional episodes depending on biological and environmental, but also on socio-cultural, factors. In a nutshell, what happens in an emotional learning processes is that different sets of cognitive bases get associated with different emotional reactions. Similarly, the emphasis on cognitive bases allows the Receptive Theory to easily account for the fact that emotions can be open to so-called why questions. This is because the cognitive basis of an emotion can be considered to explain the emotion they cause. The reason for which you feel fear when seeing a bear, on this suggestion, is that you are seeing a bear. The visual experience of the bear, thus, explains why you feel fear.

I have emphasized the continuity between the Perceptual Theory and the Receptive Theory. But are there not deep differences between the two accounts? In particular, what should we conclude concerning the justificatory power of emotions? At least for some theorists, one important attraction of the Perceptual Theory is that it promised to make room for an attractive epistemological story regarding our grasp of evaluative features. Can this story be upheld within the framework of the Receptive Theory? This is the question I turn to in the next section.

4. Epistemological Upshots

An important attraction of the Perceptual Theory is that it promises to offer a solution for a tough epistemological nut to crack. Echoing John L. Mackie’s claim in his book Ethics: Inventing Right and Wrong [45], many are skeptical that one can develop a plausible epistemology of the normative. According to Mackie, there would be no way to epistemically access objective normative stuff, something that he thought cast doubt on the existence of that kind of stuff. The best explanation of this lack of epistemic access, in his mind, is simply that there is no objective normative stuff.

Now, it can be agreed, of course, that there is no special organ that has the function of detecting normative features, be they evaluative, deontic, or of some other kind. What the Perceptual Theorist can say, however, is that when it comes to evaluative properties, or at least to properties such as being fearsome, admirable, or disgusting, i.e., affective properties, there is a simple story to be told. That story is one that is exactly parallel to the story many find plausible in the case of sensory perception. Sensory perception is widely considered to play a fundamental, though defeasible, role with respect to the justification of sensory beliefs. On the basis of seeing a poppy as orange, you are prima facie justified in believing that the poppy is orange. You could, of course, be wearing orange-tinted glasses, and your belief might be false.

Now, according to the Perceptual Theory, a similar story appears plausible. That emotions are considered to be a kind of perception does not entail that they have justificatory power. Emotions might be less well-placed, compared to sensory perceptions, to serve as the basis of justification. They might be too fickle or too much under the influence of social norms. Nonetheless, the Perceptual Theory strongly suggests an epistemological story on the same lines as the sensory perception one. According to that story, emotions would play a fundamental, though defeasible, role in the justification of at least some evaluative beliefs, i.e., affective beliefs.18 In a nutshell, the epistemological claim is the following:

Minimal Epistemological Claim: Emotions can confer prima facie justification to evaluative (or, more narrowly, affective) beliefs.

This minimal claim has been quite popular.19 Apart from the support it appears to receive from the Perceptual Theory, a good reason to embrace it is the plausibility of a Sentimentalist or, more precisely, a Neo-Sentimentalist account of the evaluative.20 It is hard to deny that at least a good number of evaluative concepts are tightly connected with affective responses: think about the concept of the admirable and admiration, of the concept of the disgusting and disgust, or of the concept of the fearsome and fear. There are quite different ways to spell out the neo-sentimentalist idea, some theorists being interested in an account of properties, others in one that concerns concepts, and the accounts diverging radically in terms of their aim, some being attracted by reductive theories while other favouring non-reductive ones. However, most Neo-Sentimentalists agree that something falls under an evaluative concept such as fearsome, admirable, etc. if and only if a correlated emotional reaction, such as fear, admirable, etc., is fitting. Thus, it appears that at least these kinds of evaluative concepts have to be elucidated in terms of fitting emotional responses. Now, a plausible thought is that Neo-Sentimentalism has an epistemological root: the story that explains Neo-Sentimentalism is simply that to determine whether something is fearsome, admirable, etc., there is nothing more fundamental to appeal to than our responses of fear, admiration, etc.21

There are different ways to develop this minimal claim into a full-bodied account of justification, including sans phrase justification, by contrast to prima facie justification. Here is one way to go, which adopts defeater epistemology as its framework, illustrated with a case of fear:

(Prima Facie Justification) Your belief that x is fearsome is prima facie justified if

(a)

You feel fear towards x;

(b)

Your belief is based on your fear.22

The claim that emotions constitute defeasible reasons for at least some evaluative beliefs fits well with a foundationalist account of justification, according to which epistemic justification depends on beliefs whose justification is independent of other beliefs. However, the same claim can also be accepted within a broadly coherentist epistemological framework (see [52,53]). As Antti Kauppinen puts it, beliefs based on emotions can be taken to constitute “initially credible starting points in a process of seeking reflective equilibrium” ([53] (p. 361)). Recently, it has been argued that this approach can be defended within a reliabilist framework (see [59]) as well as within a virtue epistemologist one (see [60]). Thus, a variety of approaches can make room for the claim that emotions have a crucial epistemic function in that they constitute defeasible but essential epistemic “touchstones” for our evaluative beliefs, to use Marilyn Friedman’s striking expression [65].

Now, consider briefly the more ambitious claim about sans phrase justification that is not merely prima facie:

(Sans Phrase Justification) Your belief that x is fearsome is justified sans phrase if

(a)

You feel fear towards x;

(b)

Your belief is based on your fear;

(c)

You have no reason to believe that defeaters are at work.

What the last condition says, is that there is nothing you are aware of that suggests that your emotion is misleading you. Put differently, you have no reason to believe that your fear is incorrect. Defeaters, in this context, are objective conditions that tend to distort your emotions, so that when a defeater is at work, your emotion is likely to be incorrect.23 Being the butt of a joke, being tipsy, having smoked pot, but also disagreeing with a peer, are good examples of defeaters. It is important to notice that the set of conditions is fallibilist. The satisfaction of these conditions does not entail the truth of your belief.

For completeness, let me briefly mention what conditions of knowledge could be envisaged, this time illustrated with the case of admirableness:

(Knowledge) S knows that x is admirable if

(a)

S believes that x is admirable;

(b)

S’s belief that x is admirable is justified sans phrase;

(c)

There are no ultimately undefeated defeaters of the justification of your belief that x is admirable.24

Whatever the exact way to spell out both the justification conditions, let it be granted that they have some initial plausibility. The question here is not whether such justification conditions are acceptable. Rather, I wish to discuss whether or not adopting the Receptive Theory is compatible with subscribing to the same epistemological story.

The answer, I believe, is that you can well embrace the Receptive Theory without having to discard the epistemology that has been associated with the Perceptual Theory. The gap between the two approaches is not as wide as to make it impossible to hold that emotions play a central epistemic role in the justification of affective beliefs. This should not come as a surprise, for the central difference between the Receptive Theory and the Perceptual Theory does not concern the characteristics of emotions, but the question of what counts as a perceptual state. Both the Receptive and the Perceptual Theory agree on the list of five characteristics of emotion we started with: that emotions typically have phenomenal properties; that they typically involve motivation; that they typically have intentional objects; that they typically involve appraisals; and that they typically have formal objects. But more specifically, the two theories agree that emotions are representational states, which have correctness conditions involving affective properties, analog format and non-conceptual contents, and cognitive bases on which they depend.

Furthermore, the arguments in favour of the epistemological claims based on the Perceptual Theory can easily be used by an advocate of the Receptive Theory. First, the Receptive Theory, by holding that emotions are representations concerning affective properties is well-placed to lend support to the epistemological claims. That it is hard to deny that evaluative concepts of the affective kind are tightly connected to affective responses is something that speaks for the epistemological claims quite independently of whether the Perceptual or the Receptive Theory is true. Put otherwise, the insight that there is nothing more fundamental to appeal to than our emotional responses in order to determine whether something has an affective property is entirely neutral as to whether emotions have to be conceived of as perceptions or as analog representations that lie at one remove from the sensory periphery.

Empirically minded philosophers might object that only perceptions have an immediate epistemic power. Therein, perception would differ from beliefs, which can justify other beliefs only provided that they are themselves justified. This, empirically minded philosophers might argue, is precisely because sensory perceptions are both stimulus-dependent and tied to a sensory modality. Only states that have these two characteristics, and thus lie at the periphery of cognition are liable to have immediate epistemic power.

I can think of two replies to this objection. The first one is concessive. It consists in accepting that emotions only afford mediate justification. To have any justificatory power, emotions need to be based on justified cognitive bases (see [4,76]). Unfortunately, two problems arise. As Michael Brady [5] has argued, that cognitive bases are in a position to justify emotions appears to nullify the justificatory power of emotions. The thought is that if emotions need to be justified to have justificatory power, it can be concluded that they merely transmit the justification they receive from their cognitive bases. Hence, emotions would lack proper justificatory power and be epistemically superfluous, the short route that takes us from cognitive bases to evaluative beliefs being obviously more parsimonious.25 The second problem is that cognitive bases are, as such, incapable of justifying emotions. Suppose you see a bear standing up in front of you and feel fear on the basis of this. That you see a bear standing up in front of you might give you reasons to feel fear—feeling fear surely makes sense in this situation—but your perception fails to justify your fear. What would seem necessary, in addition, is that you believe that the bear you see is a threat to yourself, something that would seem to require that you believe that bears standing up in front of you are dangerous.

Let us consider the second reply. It consists in denying that only states located at the periphery of cognition can afford immediate justification. If so, emotions would not need to be stimulus-dependent and tied to a sensory modality to have justificatory power. To bolster this reply, it would be good to be able to mention other non-peripheric states that have justificatory power. This is a tentative thought, but it seems reasonable to hold that analog magnitude representations have justificatory power of the same kind as the one I take emotions to have. If pigeons were able to form conceptual representations of numbers, they would do so on the basis of their analog number representations. And indeed, it is plausible to assume that our own, human numerical beliefs are not only often based on our analog number representations, but that such beliefs are prima facie justified by these analog number representations.26 Overall, there appears to be no reason to deny that analog representations can have justificatory power.

5. Conclusions

What has emerged is a novel theory of emotions, the Receptive Theory, that is close enough to the Perceptual Theory to share its main virtues without being embarrassed by what has often been seen as its main defects, that is, it is being committed to the claim that emotions are, quite literally, perceptions of a certain kind. As we have seen, this claim is difficult to believe if one is attached to a strict, non-liberal notion of perception. According to the Receptive Theory, emotions are analog representations that non-conceptually represent their intentional objects as having properties such as being fearsome, disgusting or admirable, and which lie at one remove from the sensory periphery in that they are stimulus-independent and amodal.

Importantly, the Receptive Theory can uphold the epistemological claim that has been associated with the Perceptual Theory. There is just as much reason to claim that emotions can confer prima facie justification to evaluative (or more precisely affective) beliefs when you embrace the Receptive Theory as when you defend the Perceptual Theory. Moreover, there is little reason to hold that only perceptions have immediate justificatory power. This is good news for the quest of an epistemology of the evaluative.

To wrap up, let me state a set of remaining worries, which have given rise to objections against the claim that emotions have justificatory power (see [5]). As we have seen, one worry with the epistemology under consideration is that insofar as emotions are open to why questions, they cannot serve as regress-stoppers. Another worry is that emotions are much too unreliable and idiosyncratic to serve as epistemic grounds. A further worry is that given the attentional focus that they induce, it appears that emotions’ function is to raise questions, and not to answer evaluative questions. One might also worry that if you suppose that emotions have justificatory power, then you are committed to the claim that an emotion can justify itself, a doubtful feat. This seems to be the case if you accept that emotion can be justified by an evaluative belief, such as the belief that there is danger, a belief which, it would seem, is exactly the type of belief to which your fear could provide prima facie justification. Finally, one can worry that if Neo-Sentimentalism is correct, and so the evaluative can be spelled out in terms of fitting responses, then again there is a threat of self-justification. This is because fear, for example, could justify the belief that something makes fear fitting, and so one would have to conclude that fear can justify itself.

Should we not prefer to ground our evaluative beliefs on reasoning or on rational intuitions, as the Rationalists among us would argue? I would hope not. Independently of where your preference stands, however, the relevant question here is where the Receptive Theory and the epistemology that comes with it stand relative to these worries. One would like to know, in particular, whether or not the Receptive Theory can help to deal with these worries. The good news, in conclusion, is that there is room for more work here.