Rodrigo Díaz

Most researchers agree that emotions have two core features. First, emotions have intentionality. That is, emotions are about something. Second, emotions have phenomenology. In other words, emotions are felt. It is commonly assumed that emotional intentionality and phenomenology cannot be separated into two distinct emotional states. Against this, we propose to divide emotions into (non-phenomenal) emotional attitudes and (non-intentional) emotional feelings. We argue that this proposal is theoretically fruitful and descriptively adequate. Its theoretical fruitfulness includes explaining away the challenge of merging emotional intentionality and phenomenology, and providing a straightforward account of emotional states that lack intentionality (such as moods) or phenomenology (such as unconscious and standing emotions). Its descriptive adequacy is supported by three new empirical studies, which suggest that people distinguish to some extent emotional intentionality and phenomenology by using different verbs: to be afraid/angry/etc. vs. to feel afraid/angry/etc.

Are emotions bodily feelings or evaluative cognitions? What is happiness, pain, or “being moved”? Are there basic emotions? In this chapter, I review extant empirical work concerning these and related questions in the philosophy of emotion. This will include both (1) studies investigating people’s emotional experiences and (2) studies investigating people’s use of emotion concepts in hypothetical cases. Overall, this review will show the potential of using empirical research methods to inform philosophical questions regarding emotion.

One of metaepistemology’s most central debates revolves around the question of what the source of epistemic normativity is. Epistemic instrumentalism claims that epistemic normativity is a species of means-ends normativity. One of the most prominent objections against epistemic instrumentalism features cases of epistemic indifference: Cases where there’s evidence that p yet believing that p wouldn’t promote any of the agent’s aims, wants, or needs. Still, there’s an epistemic reason for the agent to believe that p and thus epistemic instrumentalism is false. In response, instrumentalists have modified their views in various ways, with new contributions still forthcoming. Here, we investigate a neglected aspect of this debate: Laypeople’s judgments on cases of epistemic indifference. In two studies, we investigated whether laypeople agree with the verdict in cases of epistemic indifference as well as the key ideas behind the more recent instrumentalist replies. Our findings indicate that a significant amount of participants found it hard to “buy into” the cases of epistemic indifference as Kelly has constructed them. Participants did generally share Kelly’s judgment in cases of epistemic indifference. Lastly, some instrumentalist replies are well suited to explain participants’ judgments that agents ought to believe in cases of epistemic indifference.

It is often claimed that fear has an important epistemological function in making us aware of danger. Reactive theories challenge this view. According to them, fear is a response to real or apparent danger. In other words, real or apparent danger is the reason for which we experience fear. Thus, fear depends on awareness of danger instead of making us aware of danger. Proponents of the reactive theory have appealed to phenomenological and, most prominently, linguistic observations to support their views. In particular, they argue that how we talk about the objects of fear supports the view that fear is a response (rather than a perception), and that how we talk about reasons for fear supports the view that reasons for fear are necessarily evaluative (about danger). Building on systematic linguistic observations in the form of corpus analyses and survey data, we argue that how we talk about the objects of fear and how we talk about reasons for fear do not support the reactive theory. Indeed, these linguistic considerations undermine the reactive theory. Most importantly, they suggest that there are non-evaluative reasons for fear. This is good news for rival theories, like perceptual theories, which hold that emotions have epistemic value in contributing to knowledge of evaluative properties such as danger.

Human emotional expressions can communicate the emotional state of the expresser, but they can also communicate appeals to perceivers. For example, sadness expressions such as crying request perceivers to aid and support, and anger expressions such as shouting urge perceivers to back off. Some contemporary artificial intelligence (AI) systems can mimic human emotional expressions in a (more or less) realistic way, and they are progressively being integrated into our daily lives. How should we respond to them? Do we have reasons to reply to the appeals made by AI emotional expressions? In this paper, we examine the conditions under which AI emotional expressions could give us prudential or even moral reasons to change our behavior. We argue that these conditions do not depend on whether the emotional expression is genuine or not, but rather on the presence of features some of which can be implemented in emotive AI given our current level of technological development. We extract recommendations and warnings for the development of emotive AI.

This work shows how systematically studying people’s use of emotion concepts (what people think emotions are), can inform debates regarding the nature of emotion (what emotions are). As such, it makes a contribution both in terms of method and content. In regards to the methodological approach, this work constitutes the first experimental philosophy Intuitions Project approach (see Article 4) to general questions regarding the nature of emotion. It does not only bring together the philosophical and scientific literature on emotion (as much research has already done) but directly combines philosophical analysis and empirical research methods to study emotion. Articles 1 – 3 use quantitative (inferential statistics) and qualitative (text analysis) methods on data from surveys whose design was guided by philosophical analysis. And Article 4 considers the use of neuroimaging methods in philosophy. Finally, Article 2 builds bridges between analytic philosophy and the scientific study of concepts. In particular, by considering concepts’ internal structure (typicality, weight, and dependency relations). In regards to the content, the main contribution of this work consists in testing the premises used in arguments for (and against) emotion theories. This helps us discriminate between sound and unsound arguments, and select the appropriate conclusions regarding the nature of emotion. Figure 1 graphically represents the state of the debate after this contribution (see Figure 5 in the Introduction for a comparison point). In the following, I will detail the contribution of each of the articles in this work, and how they converge and complement each other to provide a coherent picture of emotion (concepts). Figure 1. Arguments from conditional and problem cases as applicable for both Cognitive / Perceptual theories and Somatic / Attitudinal theories. Solid black arrows indicate support, dashed red arrows indicate opposition. Claims challenged in this work are flagged using dashed lines. Claims not tested are flagged using transparency. The studies presented in Article 1 tested the Subtraction Argument in favor of Somatic theories of emotion. The results of these studies showed that people apply emotion concepts (fear, anger, sadness) in cases of absence of bodily feelings. This questions the claim that bodily perceptions (attitudes) are necessary for emotion (see Introduction §4.1.) and supports the existence of “calm” cases of emotion (see Introduction §4.3.2.). As we saw in the introduction, these two claims are incompatible. The first one supports the Subtraction Argument in favor of Somatic (Attitudinal) theories, while the second one supports arguments against them. With these results in hand, we can conclude that the Subtraction Argument is unsound (its main premise is likely false), and the argument from calm emotions is sound (see Figure 1). This poses a problem for Somatic and Attitudinal theories of emotion, but also Hybrid theories of emotion, as they are also committed to the claim that felt bodily changes constitute emotions. The studies presented in Article 2 went beyond those in Article 1 by testing people’s use of emotion concepts in cases where the presence (absence) of bodily feelings, cognitive evaluations, and action tendencies is independently manipulated. The results show that people consider that bodily feelings are non-necessary for emotion, in line with the results in Article 1. Action tendencies were also regarded as non-necessary, which poses a problem for Motivational theories of emotion (see Introduction §5.2.1.). Finally, cognitive evaluations were considered non-necessary in two out of three studies. However, across all studies, cognitive evaluations were the most important factor in determining people’s use of emotion concepts. Thus, it seems like we should not give up the claim that “When someone does not [value perception/belief], she is not emotion]” (see Figure 1). Furthermore, the results show that bodily changes and action tendencies depend on cognitive evaluations. These last two results square well with the core tenets of Cognitive, Perceptual, and Appraisal theories of emotion. Thus, the results presented in Article 2 support these theories. Article 3 complements Article 1 and 2 by testing the weight of yet another factor in people’s use of emotion concepts. Namely, the value of the situation in which the emotion is experienced. The results of Article 3 show that this factor significantly influences people’s use of emotion concepts. This poses a challenge for theories of emotion as non-intentional states (Morag, 2017; Whiting, 2011). According to these accounts, emotions are only causally related to the situation in which they are experienced, but they are not about that situation. Thus, they seem unable to explain why differences in the value of the situation influence the use of emotion concepts even when the reaction of the agent is the same. By stressing the importance of the intentionality of emotion and the (normative) connection between emotion and value, these results (as those in Articles 1 and 2) favor Cognitive, Perceptual, and Appraisal theories. As we saw in the Introduction, these theories are especially well-suited to explain the intentionality of emotions and their relation to value. Finally, the main goal of Article 4 is to analyze the prospects of using fMRI to inform philosophically relevant questions concerning emotion. First, we saw that there are no brain areas specifically associated with emotion. The brain correlates of emotions are also involved in other psychological processes, so we cannot infer emotion from localized brain activity. Second, and against the predictions of Basic Emotion theories (see Introduction §5.2.), we saw that patterns of brain activity are not significantly different across emotion types. This provides evidence against this particular Componential account. However, other Componential theories (e.g. Appraisal theories) are compatible with the results. Note that the same approach used to test the claims of Basic Emotion theories (Ontology testing) could potentially be used to inform questions regarding the role of evaluative cognitions / bodily feelings in emotion. In particular, we could test the degree of overlap between the neural correlates of emotion and evaluative cognitions vs. the degree of neural overlap between emotion and bodily feelings. Future research should consider this approach to further investigate what constitutes emotions. To sum up, Articles 1 to 4 challenge Somatic, Attitudinal, Hybrid, Motivational, and Basic Emotion theories, and instead favor Cognitive, Perceptual, and Appraisal theories. But before declaring winners, some limitations should be taken into account. First, it is important to note that Cognitive, Perceptual, and Appraisal theories still face challenges. Cognitive theories have to account for the emotions of non-human animals as well as the phenomenology of emotion. Appraisal theories have to account for the problem of integration. And all Cognitive, perceptual, and appraisal theories have to account for cases of recalcitrant emotion. The work presented here provides an independent motivation to develop and defend such accounts. Second, as it has been noted in the discussion of Article 1, one could account for mismatches between theoretical and ordinary uses of emotion concepts by appealing to factors that are orthogonal to emotion. For example, it might be that people’s use of emotion concepts is heavily influenced by pragmatic factors. Thus, although the results presented in this work demand proponents of Somatic, Attitudinal, Hybrid, and Motivational theories to develop and empirically support such hypotheses, it doesn’t preclude the possibility to protect these theories from the evidence. Third, note that important issues around the nature of emotion, such as those concerning emotions’ representational (see, e.g. Grzankowski, 2016) or reason-responsive nature (see, e.g. Maguire, 2017), were not treated in this work. Arguably, these questions are better addressed using philosophical analysis alone. However, future work should consider the prospects of using empirical research methods to illuminate these issues. In general, future work should continue to explore the possibilities of combining philosophical analysis and empirical research methods to inform questions regarding the nature of emotion, so we can further advance long-standing debates in emotion theory. Hopefully, this work sets a precedent for this.

Are feelings an essential part or aspect of emotion? Cases of unconscious emotion suggest that this is not the case. However, it has been claimed that unconscious emotions are better understood as either (a) emotions that are phenomenally conscious but not reflectively conscious, or (b) dispositions to have emotions rather than emotions proper. Here, I argue that these ways of accounting for unconscious emotions are inadequate, and propose a view of emotions as non-phenomenal attitudes that regard their contents as relevant to one's motivations.

Do moral beliefs motivate action? To answer this question, extant arguments have considered hypothetical cases of association (dissociation) between agents’ moral beliefs and actions. In this paper, I argue that this approach can be improved by studying people’s actual moral beliefs and actions using empirical research methods. I present three new studies showing that, when the stakes are high, associations between participants’ moral beliefs and actions are actually explained by co-occurring but independent moral emotions. These findings suggest that moral beliefs themselves have little or no motivational force, supporting the Humean picture of moral motivation.

Evaluative judgments imply positive or negative regard. But there are different ways in which something can be positive or negative. How do we tell them apart? According to Evaluative Sentimentalism, different evaluations (e.g., dangerousness vs. offensiveness) are grounded on different emotions (e.g., fear vs. anger). If this is the case, evaluation differentiation requires emotional awareness. Here, we test this hypothesis by looking at alexithymia, a deficit in emotional awareness consisting of problems identifying, describing, and thinking about emotions. The results of Study 1 suggest that high alexithymia is not only related to problems distinguishing emotions, but also to problems distinguishing evaluations. Study 2 replicated this latter effect after controlling for individual differences in attentional impulsiveness and reflective reasoning, and found that reasoning makes an independent contribution to evaluation differentiation. These results suggest that emotional sensibilities play an irreducible role in evaluative judgment while affording a role for reasoning.

In emotion research, both conceptual analyses and empirical studies commonly rely on emotion reports. But what do people mean when they say that they are angry, afraid, joyful, etc.? Building on extant theories of emotion, this paper presents four new studies (including a pre-registered replication) measuring the weight of cognitive evaluations, bodily changes, and action tendencies in people’s use of emotion concepts. The results of these studies suggest that the presence or absence of cognitive evaluations has the largest impact on people’s emotion attributions, and that bodily changes and action tendencies are considered to depend on cognitive evaluations. Implications for theories of emotion (concepts) and the interpretation of emotion reports are discussed.

Can we experience emotion without the feeling of accelerated heartbeats, perspiration, or other changes in the body? In his paper “What is an emotion”, William James famously claimed that “if we fancy some strong emotion and then try to abstract from our consciousness of it all the feelings of its bodily symptoms, we find we have nothing left behind” (1884, p. 193). Thus, bodily changes are essential to emotion. This is known as the Subtraction Argument. The Subtraction Argument is still used to support Embodied theories of emotion. Proponents of Disembodied theories, however, have different intuitions regarding the role of bodily changes in the experience of emotion. In order to advance this debate, I conducted a series of studies testing people’s intuitions regarding the Subtraction Argument. Results show that, against James’ claims, most people consider that their hypothetical (Study 1) and actual (Study 2) emotions persist in the absence of bodily feelings. Furthermore, Study 1 found that participants’ responses were not related to individual differences in reflection or interoceptive awareness. These results pose a problem for Embodied theories of emotion and challenge extant explanations for the persistence of emotional experience in cases of impaired interoception.

Emergency situations require individuals to make important changes in their behavior. In the case of the COVID-19 pandemic, official recommendations to avoid the spread of the virus include costly behaviors such as self-quarantining or drastically diminishing social contacts. Compliance (or lack thereof) with these recommendations is a controversial and divisive topic, and lay hypotheses abound regarding what underlies this divide. This paper investigates which cognitive, moral, and emotional traits separate people who comply with official recommendations from those who don't. In four studies (three pre-registered) on both U.S. and French samples, we found that individuals' self-reported compliance with official recommendations during the COVID-19 pandemic was partly driven by individual differences in moral values, disgust sensitivity, and psychological reactance. We discuss the limitations of our studies and suggest possible applications in the context of health communication.

In a recent paper in this journal, David Chalmers introduced the meta-problem of consciousness as “the problem of explaining why we think consciousness poses a hard problem” (Chalmers, 2018, p. 6). A solution to the meta-problem could shed light on the hard problem of consciousness. In particular, it would be relevant to elucidate whether people’s problem intuitions (i.e. intuitions holding that conscious experience cannot be reduced to physical processes) are driven by factors related to the nature of consciousness, or rather by factors that are independent of consciousness. Whether people hold problem intuitions, and what factors drive those intuitions, are largely empirical questions. However, there is a lack of empirical research on these issues to date. The results of four studies will show that (1) problem intuitions are not widespread, and (2) when they arise, they do so because of factors that are unrelated to the nature of consciousness. This suggests that consciousness is, after all, not so problematic.

It is generally assumed that emotion concepts are purely descriptive. However, recent investigations suggest that the concept of happiness includes information about the morality of the agent's life. In this study, we argue that normative influences on emotion concepts are not restricted to happiness and are not about moral norms. In a series of studies, we show that emotion attribution is influenced by whether the agent's psychological and bodily states fit the situation in which they are experienced. People consider that emotions are not just about feeling in certain ways, but also about feeling the right way.

Testimonial injustices occur when individuals from particular social groups are systematically and persistently given less credibility in their claims merely because of their group identity. Recent “pluralistic” approaches to folk psychology, by taking into account the role of stereotypes in how we understand others, have the power to explain how and why cases of testimonial injustice occur. If how we make sense of others’ behavior depends on assumptions about how individuals from certain groups think and act, this can explain why speakers are given different degrees of credibility depending on their group identity. For example, if people assume that women are more emotional than men, they will systematically give less credibility to women’s claims. This explanation involves three empirical claims: people assume that women are more emotional than men, people assume that emotionality hinders credibility, and people give less credibility to women’s claims. While extant studies provide some support for and, no study to date has directly tested. In two different studies, we tested all these three claims. The results from both studies provide support for, as we found significant negative correlations between emotionality and credibility attributions. However, in contrast to what some accounts of folk psychology posit, we did not find any significant difference in people’s attributions of emotionality and credibility towards women versus men speakers. We hope that our studies here pave the way for further empirical studies testing the phenomenon of testimonial injustice in a context-sensitive way, in order to have a better understanding of the conditions in which testimonial injustices are likely to happen.

This chapter analyses the prospects of using neuroimaging methods, in particular functional magnetic resonance imaging (fMRI), for philosophical purposes. To do so, it will use two case studies from the field of emotion research: Greene et al. (2001) used fMRI to uncover the mental processes underlying moral intuitions, while Lindquist et al. (2012) used fMRI to inform the debate around the nature of a specific mental process, namely, emotion. These studies illustrate two main approaches in cognitive neuroscience: Reverse inference and ontology testing, respectively. With regards to Greene et al.’s study, the use of Neurosynth (Yarkoni 2011) will show that the available formulations of reverse inference, although viable a priori, seem to be of limited use in practice. On the other hand, the discussion of Lindquist et al.’s study will present the so far neglected potential of ontology-testing approaches to inform philosophical questions.

Research has consistently shown that people consider harmful side effects of an action more intentional than helpful side effects. This phenomenon is known as the side- effect effect (SEE), which refers to the influence of moral considerations in judgments of intentionality and other non-moral concepts. There is an ongoing debate about how to explain this asymmetric pattern of judgment and the psychological factors involved in it. It has been posited that affective reactions to agents that bring about harmful side- effects could bias intentionality attributions in these cases, explaining the asymmetric pattern of intentionality judgments that we observe in the SEE. We call this the affective bias hypothesis (ABH). Evidence for the ABH is mixed, with some findings suggesting a role for affective processes, while others suggesting that affective processes play no role in the SEE. A possible explanation for these apparently contradictory results points to affective processes involved in the SEE being confined to anger. In a series of empirical studies, we systematically measured and manipulated participants’ anger in order to test this possibility. Our findings suggest that anger play no role in intentionality judgments in SEE cases, while providing support for a non-emotional motivation to blame as a factor underlying the SEE.

Responding to recent concerns about the reliability of the published literature in psychology and other disciplines, we formed the X-Phi Replicability Project to estimate the reproducibility of experimental philosophy. Drawing on a representative sample of 40 x-phi studies published between 2003 and 2015, we enlisted 20 research teams across 8 countries to conduct a high-quality replication of each study in order to compare the results to the original published findings. We found that x-phi studies – as represented in our sample – successfully replicated about 70% of the time. We discuss possible reasons for this relatively high replication rate in the field of experimental philosophy and offer suggestions for best research practices going forward.