Emotional Feelings: Evaluative Perceptions or Position-Takings? Introduction to the Special Section

This special section of Emotion Review is devoted to philosopher Jean Moritz Müller's book The World-Directedness of Emotional Feeling: On Affect and Intentionality (Palgrave MacMillan, 2019). It consists of a précis of the book by its author, six commentaries by philosophers and psychologists, and a response by Müller. ¹

Our aims with this special section are threefold. First, we want to provide the readers of Emotion Review, particularly nonphilosophers, with a spotlight view of recent research in the philosophy of emotions. In his book, Müller (2019) criticizes what, during the past 10 years or so (see Tappolet, 2016), has become the (or at least a) dominant philosophical theory of the nature of emotions: the theory that emotions—understood as a class of conscious mental states that have an experiential quality—are evaluative perceptions, that is, sensory-perception-like apprehensions of the positive or negative value that objects and events have for us (e.g., Döring, 2007; Tappolet, 2016; see already Meinong, 1917). Müller calls this theory, which actually comprises a family of related accounts, the axiological (i.e., value-related) receptivity view of emotions. Müller then proposes an alternative theory of emotional experience, the theory of emotions as affective position-takings (PT), that seeks to accommodate the criticisms of the axiological receptivity view and claims to be more faithful to phenomenological and linguistic data.

Second, by providing a forum for the discussion of an emotion theory at the forefront of current philosophical research, we want to support the further development of emotion theory and, in the case of emotion psychology, to also encourage a renewed focus on general emotion theory (see Reisenzein, 2019, 2022a).

Third, closely related to the second aim, in planning this section, we wanted to support the interdisciplinary exchange in emotion science (see also, Reisenzein et al., 2013) by encouraging a dialogue between philosophers and psychologists aimed at clarifying the relations between the emotion theories developed in the two disciplines, improving the theories, and exploring possible paths to integration (see also Moors, 2022a; Reisenzein, 2022a). Despite an increasing number of attempts to cross disciplinary boundaries (e.g., A. W. Müller & Reisenzein, 2013; Reisenzein, 2012; Teroni, 2021), discussions are still too often limited to researchers from the same discipline: Psychologists discuss psychological theories, philosophers discuss philosophical theories. ² This is unfortunate because research in one discipline is often highly relevant to the concerns of the other. To mention but two examples: (1) Emotion theories that view emotional experiences as compound states consisting of two or more components (e.g., cognitive appraisals, bodily feedback, felt action tendencies) continue to be popular in psychology (see e.g., Moors & Scherer, 2013) but have been largely abandoned in philosophy. Psychologists of emotion should want to know why, and what the alternatives are. (2) Conversely, emotion theories that accord a central role to bodily feedback for emotional experience continue to be defended by many philosophers (see e.g., Deonna & Teroni, 2020; though contrast Müller, 2019; Roberts, 2003) despite the fact that empirical evidence accumulated in psychology fails to support, and in fact even undermines strong claims regarding the importance of bodily feedback for emotion (see e.g., Reisenzein & Stephan, 2014). Philosophers should want to know this research and let it inform their theorizing.

In this special section, the two psychological commentators (Reisenzein, Moors) add to the interdisciplinary exchange by investigating how PT relates to psychological theories of emotion (appraisal theories, and a computational model of the belief-desire theory), while the four philosophical commentators (de Sousa, Furtak, Mulligan, Schmidt) shed light on PT from different philosophical traditions (analytical, phenomenological, and existential philosophy).

While different in focus, this special section of Emotion Review can be seen as an update to the special issue Ten perspectives on emotional experience, published a decade ago in the same journal (Reisenzein & Döring, 2009). Whereas that special issue featured ten articles written by psychologists and philosophers on recent theories of emotional experience, this special section focuses on a single philosophical theory, PT. However, both the target article and the commentaries contain information about the theoretical and historical context of the theory, thereby shedding light on related theories and debates in both disciplines. In addition, we will provide additional context by giving a brief review of the main types of theories of emotional experience below. This review focuses on philosophical theories of emotion but is written for the interdisciplinary readership of Emotion Review: Its aim is to allow emotion researchers from other disciplines to understand how philosophers of emotion have arrived at the current point in their debates.

On the Relation between the Philosophy and Psychology of Emotion

We want to begin with some general metatheoretical comments on the commonalities and differences between the psychology and the philosophy of emotion, and the potential relevance of one discipline for the other. The motivation for these comments is that these issues are often misunderstood, particularly by psychologists. As a consequence, many psychologists have come to believe that theories and debates in the philosophy of mind are irrelevant to their concerns. Put bluntly, many psychologists, particularly hard-nosed empiricists, consider research by philosophers of mind to be a continuation of the speculative arm-chair theorizing that, they have been taught, was characteristic of a bygone era of psychology that has been superseded by empirical psychological science. (To be sure, there are also philosophers of mind who believe that psychological research is irrelevant to their questions).

In general, scientific disciplines can differ from each other with respect to (a) their object, or domain of inquiry, (b) the specific questions asked about the object of inquiry, and (c) the methods used to obtain answers to these questions. In the case of the psychology and philosophy of emotions, the pre-theoretical object of inquiry is essentially the same: the emotions of humans—states like happiness and sadness, fear and hope, pity and joy for another, pride and shame and others sufficiently similar to these (see e.g., Ortony et al., 1988). Also, many of the questions asked about emotions by philosophers and psychologists are identical, or at least very similar (see e.g., Scarantino & de Sousa, 2018; Reisenzein, 2015): What kinds of mental states are emotions; how do emotions come into being; how are they related to cognition and motivation; what are their functions.

But there are also questions about emotions that are primarily discussed by philosophers, such as: How are emotions related to reasons and what are the criteria for judging emotions as justified vs. unjustified, appropriate vs. inappropriate; what is the role of emotions in the moral life (see Deonna & Teroni, 2012; Roberts, 2013). Although these questions are regarded by philosophers as important for the understanding of emotions, they play a comparatively small role in psychological discussions. Even the intentionality (object-directedness) of emotions, a feature of emotions that is of central concern to most philosophers of emotion (see below and Müller, 2019), plays only a subordinate role in psychological discussions, at least under this name (it is however indirectly addressed in, for example, research on the misattribution of the causes of feelings; see e.g., Greifeneder et al., 2011; Reisenzein, 2017).

Conversely, there are questions that are primarily investigated by emotion psychologists, such as: What are the computational mechanisms underlying emotions (see the commentaries by Moors and Reisenzein), which physiological reactions are associated with emotions, how and when do emotions show in the face, what are effective cures for undesired emotions such as phobias or depression.

However, the main difference between the philosophy and the psychology of emotions concerns the methods used to obtain information about their object of inquiry. Characteristic for today's psychology is the use of empirical methods, understood as systematic experiments and systematic observations with groups of participants that, with rare exceptions, do not include the investigator. Furthermore, psychologists typically use standardized measurement methods and do not restrict their data sources to first-person reports of experience, but include nonverbal and physiological measurements such as the observation of facial expressions, the measurement of physiological reactions, reaction times, evoked potentials of the EEG, and brain imaging methods (fMRI, PET).

The “data–gathering” methods of the philosophy of mind, in contrast, comprise roughly speaking two approaches. The first method is often called “conceptual analysis” and has been traditionally understood as the analysis of the meanings of ordinary language concepts—in the present case, concepts for emotions and related psychological states such as beliefs and desires (see e.g., Hanna, 1998; Hoche & Strube, 1985). However, this method is better described as the attempt to explicate, by studying the application conditions of emotion predicates and related psychological predicates, aspects of assumptions about emotions implicit in common-sense psychology (see Heider, 1958).

The second main method of the philosophy of mind is a method often called “phenomenological analysis” (e.g., Arnold, 1960). This method exists in several varieties that can be traced back either to Brentano’s (1874) idea of a “descriptive psychology”, or to Husserl's (1913/2022) “phenomenology” (see also Schmidt, 2018). In the form phenomenological analysis is probably most often used in practice (e.g., Arnold, 1960; Müller, 2019), it can be described as the detailed introspective investigation of (in the case of the philosophy of emotions) actual, remembered and most often, hypothetical cases of emotion episodes. Müller (2019) describes it as an investigation “of the way emotional feelings present themselves from a first-person point of view” (p. 10).

Seen from the psychological perspective, the first method (conceptual analysis) taps primarily the implicit knowledge contained in folk psychology, whereas the second method relies on, apart from introspection, the ability of humans to engage their mental mechanisms in “offline mode” for the purpose of simulation. Thought experiments on, respectively, what one would think, feel or do in a situation (or in situation A versus B), or which linguistic descriptions of a person's reactions would be appropriate or not in a situation (or in situation A versus B), are widely used in both phenomenology and conceptual analysis (for some examples see Müller, 2019; Reisenzein, 2022a).

While Müller (2019) combines both methods in his investigation of emotions, it should be noted that many contemporary philosophers of mind remain hesitant about phenomenological analyses (see e.g., de Sousa, 2022; and also see Müller's response). Psychologists are typically skeptical of both described methods. However, in fact, both methods are also used by psychologists: Theoretical psychologists make extensive use of them, and empirical psychologists often rely on folk psychological or phenomenological intuitions to generate hypotheses and as a preliminary or adjunct method to evaluate the plausibility of psychological claims (Reisenzein, 2022a). However, for the strict testing of theories, phenomenological and conceptual intuitions are regarded as being of limited value by most empirical psychologists; for this purpose, formal empirical methods are considered the gold standard. In addition, many emotion psychologists consider behavioral and physiological data not only as additional sources of information to complement introspection-based self-reports (as Wundt, 1896, still saw it), but actually prefer these data because they refer to intersubjectively observable phenomena and are therefore easier to verify.

The skepticism of psychologists towards the validity of phenomenological and conceptual analyses is in some cases well-justified; but in other cases, it is not. Some of the phenomena of emotion science that philosophical theories seek to account for—such as that emotions have an experiential quality including an intensity, or that emotions at least appear to be directed at objects (see below)—can be regarded as phenomenological facts and are treated as such in common-sense psychology. The same is true for many other assumptions of common-sense psychology (see Reisenzein, 2022a). Furthermore, extreme skepticism towards the phenomenological method and the analysis of folk psychology does not fit well with the fact that mental simulations and/or the interrogation of folk psychology are implicitly presupposed as valid methods by those (many) emotion psychologists who test their theories using the popular “hypothetical scenario” method (Reisenzein, 2022a; Schmidt, 2018). The results of conceptual and phenomenological analyses offered by philosophers can be checked by the same method, as the movement of “experimental philosophy” in fact proposes to do (see Knobe & Nichols, 2017). Even without that, one philosopher's thought experiments and the conclusions he or she draws from them are checked by evoking assent or dissent in others involved in a philosophical debate. The debate about emotions featured in this special section is a good example of this.

In addition to using conceptual and phenomenological analyses as sources of insight into the mind, philosophers also engage extensively in theory and argument analysis. In this metatheoretical work, the meanings of theoretical claims are meticulously scrutinized, different meanings of a claim are distinguished, the relations of different claims to each other are clarified, and the implications (or their lack) of phenomenological or conceptual intuitions—as well as increasingly often, of empirical-psychological findings—for theories are discussed in great detail. Such metatheoretical work is, of course, also done by psychologists and is, in fact, indispensable in all sciences (Reisenzein, 2022a). However, it is fair to say that philosophers are generally better prepared through their training for this metatheoretical work, as well as more aware of its importance, than psychologists.

The Core Question of Emotion Philosophy: What is an Emotion?

Although the questions about emotions asked by philosophers and psychologists show large overlap, there are differences in emphasis even on the shared questions. Perhaps the most important difference is that the central question of many emotion philosophers is “What is an emotion?”, whereas for most emotion psychologists, this question is only one among several and less important than the question of the causal generation of emotions and their effects.

The dominant philosophical approach to answering the question of the nature of emotion has been to interpret this question as “What kind of mental state (or perhaps, mental-behavioral state, see e.g., Gordon, 1987) is an emotion?”. That is, the question of the nature of emotion is interpreted as the task of locating the place of emotions in the domain of mental state types. Accordingly, philosophical analyses of emotion typically begin with an (at least implicit) taxonomy of “accepted” mental states, that is, mental states whose existence is taken for granted, and which are better understood than the emotions themselves. They then try to show that the emotions belong to one of these classes of mental states, or are combinations of two or more of them. ³ Alternatively, if an initially proposed reductive theoretical definition of emotions of this kind does not work, or on independent phenomenological grounds, the initial taxonomy is refined (e.g., by distinguishing between different kinds of cognitions such as beliefs and appearings-as; Roberts, 2003) or it is argued that emotions are a category of mental states sui generis and therefore cannot be reduced to other mental states. Nonreductive theories of emotions go back to Brentano (1874) and have in recent years attracted renewed interest in philosophy (e.g., Deonna & Teroni, 2012; 2015; Goldie, 2000; Helm, 2001; Mitchell, 2021); Müller's (2019) theory also belongs to this group.

For each proposed candidate mental state, one then investigates how well the respective theory that identifies emotions with this state, explains accepted intrinsic and relational properties of emotions suggested by phenomenological, linguistic and increasingly often, empirical-psychological data. For many philosophers, the two central properties of emotions are phenomenality and intentionality, discussed below.

The question “What kind of mental state is an emotion?” is also of interest to many emotion psychologists (e.g., Arnold, 1960; James, 1890; Schachter, 1964; Russell, 2003; Reisenzein, 2012); in fact, William James used it as the title of the article in which he first presented his famous emotion theory (James, 1884; but of course, James was also a philosopher). However, even these psychologists usually regard the question of the nature of emotions as but one question among several. The dominant view among emotion psychologists is that a theory of emotions must first and foremost describe how emotions are causally produced, and secondly which consequences (in particular which adaptive consequences) they have. Many empirical emotion psychologists even focus entirely on the causal generation and/or the consequences of emotions, neglecting the question of what kind of mental state the emotion itself is (see Frijda & Zeelenberg, 2001). ⁴

At second sight, however, the questions “What kind of mental state is an emotion?” and “How are emotions causally generated and what effects do they have?” are not as distinct as they may seem to be. According to functionalism, a prominent theory of the nature of mental states that is the backbone of the computational theory of mind widely endorsed in experimental psychology, the nature of mental states is determined, at least in part, by their causal relations to each other and to stimuli and behavior (e.g., Block, 1982). If functionalism is right, then the question of the nature of emotions, and the question of the causes and consequences of emotions, are not really separate: Answering the first question requires to answer the second. ⁵ There is in fact an important tradition of emotion philosophy, going back to Aristotle, that combines the taxonomic-structural approach with a functionalist approach (see Loaiza, in press; Reisenzein, 2012). That is, these emotion philosophers, in trying to answer the question of what an emotion is, also propose at least a sketch of a causal emotion theory (e.g., James, 1890; modern philosophical proponents of this approach are for example Davis, 1988; Gordon, 1987; Lyons, 1980).

Two Central Explananda for Emotion Theory: Phenomenality and Intentionality

While many properties of emotions have been put forward as features that an acceptable theory of emotional experience must be able to explain (or explain away), two of them are widely regarded as central in emotion philosophy: phenomenality and intentionality (see e.g., Deonna & Teroni, 2012; Goldie, 2000; Helm, 2001; Mitchell, 2021; Montague, 2009; Reisenzein & Döring, 2009; Müller, 2019). Regarding phenomenality: When a person is happy, sad, afraid etc., there is “something that it is like” (Nagel, 1974) for the person to have the emotion, it feels a particular way to have it. Furthermore, it is widely assumed that this experiential quality is different from that of nonemotional mental states (e.g., feeling tired, or having a visual perception of an object), differs more or less between different emotions (e.g., happiness and sadness), and can be instantiated in different intensities (e.g., feeling slightly or highly angry). An acceptable theory of the nature of emotions should be able to explain these features of emotions; that is, it should allow us to understand why emotions have these phenomenal properties.

Regarding intentionality: When one is happy or sad etc., one is at least in the typical case (many theorists believe always) angry or sad about something, or at least emotions seem to present themselves that way to the experiencer. This something is the emotion's intentional object (Brentano, 1874). For example, John is afraid of the dog, or afraid that the dog will attack him. In the more recent philosophical literature, the intentional object is often also called the particular object or target of the emotion (e.g., de Sousa, 1987) to distinguish it from the so-called formal object of the emotion, ⁶ and intentionality is typically understood as mental representation. On this understanding, to say that emotions are intentional or object-directed mental states is the same as saying that they are representational states, states that represent objects to the subject.

Preview of Commentaries

Kevin Mulligan (2022) has previously proposed a view of emotions similar to PT (e.g., Mulligan, 2010); hence he is fundamentally in agreement with PT. Nevertheless, he raises a number of questions about the theory, including questions about the motivational basis of emotions, that is the cares, attachment and concerns (CAC) invoked in the evaluation of objects, and their relations to values. According to PT, emotions presuppose that the person has already come to perceive a particular thing or event as bearing positively or negatively on her CAC. While Mulligan agrees that these concern-based construals provide motivating reasons of emotions, he argues that a satisfactory account of the evaluative basis of emotion must be backed up by an adequate account of cares and concerns. Among other things, such an account needs to accommodate for the possibility that concerns, like emotions, admit of justification, and it also must tell a plausible story of their origins. Neither of theses questions are addressed by Müller. Although Müller's theory of emotion is not the only one that is silent on these question, answering them is important to show that a CAC-based account of emotions does not lead into an infinite justificatory regress or turns out to be incoherent. The first would be the case if (all) concerns, like emotions, were responses to the perception of congruence of incongruence of other concerns. The account might turn out to be incoherent if some concerns can come into being through emotions: emotions would give rise to concerns whose existence the emotion-antecedent evaluation already presupposes. A reflection on different cases of CAC suggests to Mulligan that some CAC seem indeed to have their origins in emotions. For example, falling in love with a person may give rise to an attachment to this person. The question then is if and how PT can deal with these cases.

Ronald de Sousa (2022) compares PT to his own view of emotions, noting that while he had previously (de Sousa, 1987) proposed an analogy between emotion and perception, he has meanwhile come down firmly on the side of an attitudinal view of emotions similar to PT. One important difference remains, however: Although de Sousa agrees with Müller that emotional feelings do not detect or represent value, he also rejects the assumption that they are detected prior to emotional feeling (as PT implies). According to de Sousa, values are neither detected in the emotion nor prior to the emotion simply because they are not detected at all; instead, they are conferred to objects through emotion. Related to this, the formal object of an emotion (e.g., in fear, the property of being dangerous) is not a value property, but a natural property of the object on which the emotional response confers value. Correspondingly, emotions-as-attitudes as conceptualized in de Sousa's theory are not directed at objects that have been recognized as having positive or negative value, but at things that only come to be (dis)valuable through the emotional response. In addition to this theoretical difference, de Sousa also makes several objections to Müller's argumentation against the perceptual theory of emotion and in favor of PT. For example, he argues that Müller's claim that emotions cannot be perceptions of value because they presuppose a preceding apprehension of value rests on an unusual understanding of apprehension as an instantaneous event. The more natural understanding is that apprehension is a temporally extended process. Furthermore, de Sousa proposes that the value of an object could be detected by a prior unconscious, subpersonal process and that the person could only be aware of the feeling elicited by the evaluation; in this case, the proponent of the evaluative perception theory might argue that the value is apprehend by the feeling.

Rick Anthony Furtak (2022) is a proponent of the evaluative perception theory of emotions (Furtak, 2018). As might therefore be expected, he is skeptical of Müller’s (2019) criticisms of the perceptual theory and the proposed alternative. Like Schmidt and Reisenzein, Furtak sees no necessary conflict between the claim that emotions disclose or reveal values, and the claim that they are best described as responses—for example, the claim that the feeling of grief over the loss of a friend discloses the negative value of this event to the grieving person, and the claim that her grief is best described as an emotional response to this value. A main reason is that Furtak is unconvinced by Müller’s (2019) claim that the direction of intentionality differs in perception and emotion—that in perception we are passive while the world speaks to us, whereas in emotion we are active in reaching out to the world by taking an evaluative position. Accordingly, Furtak is not convinced that “registering value” is passive and “position-taking” is active. Even granting that an emotion might be regarded as being in some sense a decision, Furtak argues, this does not imply that emotions are active rather than passive mental states. This is so because even if an emotion were a decision, it would be “one that we are not conscious of making and that may impose itself on us” (Furtak, 2022, this issue). Furtak also refers to cases (e.g., Sartre's reaction to the death of Camus) that seem to support the idea that sometimes, the evaluative meaning of an event that one is aware of only registers later, when one experiences the emotion, suggesting that it is indeed the emotion-as-response that makes one aware of the evaluation.

Philipp Schmidt (2022b) begins his commentary by asking how exactly one should understand Müller's claim that feelings, in contrast to what the perceptual theory of emotions assumes, are not epistemically significant. Schmidt proposes that this claim can be interpreted in different ways, including an extreme interpretation that denies an epistemic function to any kind of feeling, not just to emotional feeling. Schmidt then argues that on both the wider and narrow reading, Müller's thesis faces serious counterexamples that together show that emotional feeling can have an epistemic function, and that it can have this function even if it is understood as affective position-taking. To support this claim, Schmidt describes four classes of cases where feelings may be said to be informative, and so relevant to the project of acquiring knowledge about oneself and the world: existential feelings, such as feeling tired or exhausted, or feeling detached from the world; “emotional emanations”, that is experiences where emotions are felt before the person has acquired a clear understanding of the evaluative significance of the eliciting event; cases of recalcitrant emotions, emotions that persist even when one does not assent to their evaluative contents (see Döring, 2015); and “emotions-as-heuristics”, that is cases where people imagine emotion–evoking situations to find out how important they are for them. These cases, Schmidt argues, pose problems for Müller's theory of emotions as affective position-takings, at least as it is currently formulated.

Agnes Moors (2022b) compares Müller's theory of emotions as affective position-takings to psychological appraisal theories of emotions. As Moors points out, appraisal theories share a central assumption with both the perceptual theory of emotion criticized by Müller (2019) and with PT: the assumption that emotions essentially involve an evaluation of “stimuli” (Moors' term for the objects of evaluation). What is more, PT and some versions of the perceptual theory (e.g., Roberts, 2003) share with (most) appraisal theories the more specific assumption that the process of stimulus evaluation involves the comparison of the stimuli to the concerns (desires, goals) of the person (e.g., Frijda, 1986; Lazarus, 1991; Scherer, 2001). Comparing further assumptions of the theories reveals however that appraisal theory is closer to PT than to the perceptual view of emotions. In fact, in Moors' view, both appraisal theory and PT can be regarded as attempts to solve two central problems of the perceptual theory: The abstract concept objection (also addressed in Reisenzein's commentary), and the problem of accounting for the “affective warmth” of emotions, their special phenomenal character. According to Moors, the solutions to these problems explicitly or implicitly offered by PT are close to those offered by appraisal theory. Notwithstanding these similarities between PT and appraisal theory, Moors sees several problems with PT. Although she agrees that “taking position” is indeed central to emotional experience, she points out, like Furtak, that the view of emotions as position-taking seems to be in conflict with the intuition that emotions are passive phenomena. She also criticizes PT for remaining silent on the issue of emotional action tendencies, which she regards as those mental states that come closest to “taking a position”.

Finally, Rainer Reisenzein (2022b) compares PT with his own view of emotions, which is a computational elaboration of the belief-desire theory of emotions (CBDTE) (e.g., Reisenzein, 2009a, 2009b). Reisenzein notes that the two theories share the basic assumptions that a core subset of typical emotion episodes are the products of an evaluation process in which cognized states of affairs are evaluated for their congruence with the person's desires; and that emotions are, on the conscious level, feelings of pleasure and displeasure. However, there are at least two important differences between the theories. First, according to PT, the evaluation of the emotion-eliciting object is a concern-based construal (Roberts, 2003), whereas according to CBDTE, the evaluation is performed by a subpersonal mechanism. Second, PT assumes that emotional feelings are intrinsically object-directed, whereas CBDTE takes them to be intrinsically objectless. Reisenzein argues that conceptualizing the evaluation process as a subpersonal process is preferable because it avoids several difficulties of the concern-based construal account of evaluation, as well as earlier accounts that appeal to evaluative beliefs (Reisenzein, 2009a); in particular the problem that the evaluation process as conceptualized in these theories is too cognitively demanding to be realistic. As regards the assumption of PT that emotional feelings are intrinsically intentional, Reisenzein objects that this is not a datum of introspection but a theoretical postulate; and this postulate, Reisenzein claims, is wrong. Finally, Reisenzein argues that in the framework of CBDTE, the view that emotions are perceptions of value is actually compatible with the view that they are responses to detected value (PT), because hedonic feelings are both: reactions to the (albeit subpersonal) detection of value, and nonconceptual representations of the detected value.

Müller responds to the commentaries in his reply.