Abstract
Emotions in scientific research are typically portrayed as short-lived responses or dispositions to manifest such responses. Some philosophers have argued that this fails to capture long-term emotions (e.g., love, hate, and grief). This article examines whether the emerging field of affect dynamics (or emotion dynamics), which studies how emotions fluctuate over time, can address the philosophical critique. I argue that there are still aspects of long-term emotion (i.e., their temporal components and temporal dynamics) missing from affect dynamics. I end by proposing a few positive steps psychologists working in affect dynamics can take to mitigate these shortcomings.
Introduction
Do we have a fully fledged psychological science of long-term emotions? The criticism that empirical psychology doesn’t account for such phenomena has typically come from philosophers working on emotion (Goldie, 2000; Ratcliffe, 2023; Solomon, 2007). Emotions in scientific research are treated as either short-lived responses, in other words, occurrent episodes, or as dispositions to manifest such brief responses (Ekman, 1973; Frijda et al., 1991; Panksepp, 1998). The philosophical critique is that long-term emotions are more than that. What they, in effect, are varies from philosopher to philosopher, but the core criticism remains the same: a science that reduces emotions along these lines fails to capture the way some emotions involve long-term phenomena.
That scientific research doesn’t fully capture long-term emotions is also increasingly recognized by psychologists themselves. This is especially the case for those working in the emerging field of affect dynamics (or emotion dynamics), which aims to study the way emotions fluctuate over time (Kuppens & Verduyn, 2017; Verduyn et al., 2015; Waugh & Kuppens, 2021). The question this article aims to investigate is whether affect dynamics can address the philosophical critique.
The article is structured as follows. First I explain the philosophical criticism of scientific emotion research, and hone in on what I take to be the main criticism (Philosophical critiques of affective science section). I then summarize recent developments in the field of affect dynamics and explain how they address some of the philosophical worries (The field of affect dynamics section). Following this, I identify what is missing from this field (The temporal components and dynamics of long-term emotions section). Finally, I propose a few positive steps psychologists working in affect dynamics can take in order to address these shortcomings (Towards an integrative approach to long-term emotions section).
Philosophical Critiques of Affective Science
Empirical emotion research operationalizes emotion as a coordinated pattern of expressive, physiological, behavioral, cognitive, and phenomenological responses, triggered by an object or event of personal significance (APA, 2018). Crucially, such responses are said to last seconds or minutes (Ekman, 1973, 2004). This way of conceptualizing emotion has come under criticism by philosophers who argue that emotions are more than brief episodes (Goldie, 2000; Ratcliffe, 2023; Solomon, 2007). Some caveats will prove helpful here.
First, sometimes the criticism is leveled at a particular empirical theory of emotion. For example, Solomon (2007, p. 19) argues that “anger is much more than a basic emotion or a set of feelings. It is a way of interacting with another person (or with a situation or a task) and a way of situating oneself in the world.” Solomon's critique is aimed at basic emotion theories, which propose that there is a set of short-lived affective responses that are found across most, if not all, cultures (Ekman, 1973; Izard, 1977; Panksepp, 1998). However, it is clear that the criticism applies to any theory that takes emotions to be brief episodes. As Solomon (2006, p. 303) notes, emotions “can be complex and long term—lasting even for years or a lifetime.”
Second, the criticism is also often made in relation to some positive theory of what emotions actually are. For example, philosophers working in the phenomenological tradition often emphasize the social, situated nature of our emotional responses. This is apparent in Solomon's work, which in essence treats emotions as “engagements with the world” (2006, p. 300). We also see this in more traditional phenomenologists, like Merleau-Ponty (1964, p. 53) who argues that emotions are “a variation in our relations with others and the world.” What concerns us is not whether phenomenologists’ (often poetic) characterizations of emotion are right, but what they suggest is missing from scientific emotion research. As before, the issue seems to be long-term emotions. This is clear in Ratcliffe's (2023, p. 39) recent work on the phenomenology of grief (bereavement) where he observes that “Grief involves recognizing and responding to a disturbance of one's world, something that takes time.” 1
Third, on occasion, the criticism comes not from a particular philosophical theory of emotion per se, but a broader account of emotion ontology. In brief, emotions can either be treated as processes, that is, occurrent emotional episodes (Goldie, 2011; Robinson, 2017), or dispositional states, in particular, dispositions to manifest such emotional episodes (Naar, 2022; Stout, 2022). Treating emotions as dispositional states helps recognize their long-term nature: while emotional episodes are brief, dispositional states are longstanding. For example, an episode of fear, say toward a dog, might only last a few seconds or minutes, but a disposition to fear dogs might last an entire lifetime. On this way of construing emotions, a science of emotion neglects long-term emotion because it assumes the wrong sort of emotion ontology; it focusses on emotional processes and not emotional dispositions.
However, it is worth noting that the same objection can, and has, been made by those who subscribe to the process view of emotion. For example, according to Goldie (2011, 2012), emotions are processes, but they are much too complex to be treated merely as brief episodes: [T]he emotion of grief is a kind of process – a complex pattern of activity and passivity, inner and outer, which unfolds over time, and the unfolding pattern over time is explanatorily prior to what is the case at any particular moment, and moreover, explanatorily prior to any particular mental state or event at any particular moment which is part of the process. (Goldie, 2011, p. 136–137)
In summary, the critique that scientific emotion research overlooks long-term emotions often arises within philosophical discussions concerning the nature of emotion. However, it would be a mistake to think that the criticism can only be made by making certain philosophical commitments. As we see, the criticism can be made in a theory-neutral manner, which underscores the need for more cross-disciplinary engagement with this issue.
Before we examine what the emerging field of affect dynamics holds for such worries, let me mention two obvious responses. The first response to such worries is to adopt a dispositional account of long-term emotions. As noted, we find such views in philosophy, where emotions are seen to be dispositions (Naar, 2022; Stout, 2022). For example, according to Wollheim (1999, p. 8), “if one thing is clear about the emotions, it is that they are mental dispositions.” Similar views have also been proposed in psychology with respect to sentiments (Frijda, 1986; Scherer, 1984). To clarify, within psychology, unlike in philosophy, sentiments aren’t taken to be emotions themselves, but rather dispositions to manifest emotions. The difference seems primarily semantic. However, regardless of how we understand this difference, the important point for us is that both the dispositional account of emotion and the dispositional account of sentiment seems to, prima facie, handle the philosophical critique. Put simply, since dispositions tend to be stable states that can last a long time, it looks like they can account for what we are referring to as long-term emotions.
Unfortunately, this response doesn’t appreciate the philosophical critique in its entirety. What's at issue, as the aforementioned quotes illustrate, is not just that some emotions last a longtime, but they are also complex; they are much too complex to be accounted for by brief irruptive patterns that are typically taken to constitute an emotional episode. To elaborate, anything can be given a dispositional characterization. The problem is not about whether we should describe emotions as dispositions, but how we should describe them, even if we are happy to describe them in dispositional terms. If occurrent emotional episodes themselves involve complex patterns which help situate us in the world, simply describing an emotion as a disposition to have, say, short bursts or affect program responses is to fail to account for the complexity manifest in our emotional episodes.
For example, Griffiths (2002, p. 239), who advocates an affect program account of basic emotions, acknowledges that such programs don’t account for complex emotions: “When a woman's feeling of guilt explains her behavior through a long session of negotiation with her husband and lawyers, it does more than dispose her to intermittently display affect-program sadness and affect-program fear.” Griffiths (2002), as I read him, is not denying that guilt can be given a dispositional characterization. Rather, his point is that even if guilt is a dispositional state, it is too complex a state to be characterized merely as a disposition to have certain affect program responses. This brings us back to Solomon's (2006, p. 303) main observation that “emotions can be both complex and long-term.” Simply switching to a dispositional view of emotion (or sentiment) does not fully address such observations. 2
The other obvious response is to claim that what philosophers say is missing from a science of emotion can be captured by moods. Moods, like emotions, are occurrent states that have a phenomenology. However, unlike emotions, moods tend to be long-lasting. Note: this is not to say that moods always last a long time, or that emotions are always short-lived. Emotions like grief, as stressed, can last years. By contrast, some moods are transient (Stephan, 2017). For example, you might be in irritable mood for only a brief period this morning. But moods tend to last days, months, or years, as seen from mood disorders, such as depression, so maybe research on such disorders accounts for what some philosophers claim is missing from emotion research.
The growing experimental literature on moods, however, is unlikely to directly address the philosophical critique owing to one of the main ways we demarcate moods from emotion: moods aren’t intentional in the sense that they target specific objects, whereas emotions are (Deonna & Teroni, 2012). That is, emotions are always about something, in particular, something specific. You have a fear of dogs, you love your partner, you are grieving that your parent died, and so on. By comparison, moods don’t have to be about any specific object, person or event. One of the striking things about depression, for instance, is that it is hard to articulate why it is that you are depressed. The same could be said even if one assumes a dispositional account of emotions and moods. We can say that someone is in an irritable mood if they are disposed to snap at everyone and everything, whereas we can say that this person is in love with Griselda because they are disposed to respond emotionally to her; they are disposed to feel happy when something good happens to her, feel fearful when she might come to harm, and so on. Since moods are distinct from paradigm examples of long-term emotions (e.g., love and grief) in this way, the research on moods won’t provide a story of long-term emotions as such.
The Field of Affect Dynamics
That emotion research in empirical psychology neglects long-term emotions is also recognized within the field itself (Barrett, 2014; Mesquita & Boiger, 2014). As Kuppens and Verduyn note: [F]or pragmatic reasons, emotion research has for long largely neglected the time dynamic aspects of emotions, mostly studying emotions as stable traits, or as brief states that simply switch on, stay, and switch back off (like a lightbulb) in response to events or experimental manipulation. (Kuppens & Verduyn, 2017, p. 22)
Four key principles help unify this approach: the principles of contingency (emotions are responses to things extrinsic to them), inertia (emotions display an intrinsic resistance to change), regulation (emotions are continuously regulated to match their fit with the desired state), and interactions (emotions and their components continuously interact over time; Kuppens & Verduyn, 2015; Verduyn, 2021). Methodological advances have also made the application of such principles more effective and readily available. For example, the rise of experience sampling, where people's emotions are measured not via retrospective reports but in the moment, often using wearable tools, helps avoid recall bias. It also helps capture emotions as they unfold naturally in everyday life, rather than in manipulated laboratory settings (Pirla et al., 2023). 4
The field of affect dynamics, which employs such principles and methods, has the potential to make great strides in capturing the temporal dimensions hitherto missing in empirical emotion research. First, emotions are multicomponent phenomena, consisting of a range of responses, such as expressive, physiological, behavioral, phenomenological, and so on (Frijda et al., 1991; Scherer, 1984). Affect dynamics aims to investigate the way these distinct components might have different time courses. For example, Mauss et al. (2005, p. 175) aimed to investigate the response coherence of emotion, that is, “the coordination, or association, of a person's experiential, behavioral, and physiological responses as the emotion unfolds over time.” To test this, participants were made to watch the same 5-min film which involved an amusing scene, a sad scene, and a second amusing scene, 3 times in a row. In each viewing, participants either just watched the film or were asked to continuously rate their experience of sadness or amusement using a rating dial (i.e., as they watched the film). In the first viewing, their expressive behavior was videotaped. Moreover, throughout the session, their cardiovascular activation, skin conductance level, and somatic activity were also assessed. It was found that experience, facial behavior, and peripheral physiology were significantly associated. However, results also showed that such associations were not absolute. In particular, while the intensity of amusement experience was associated with the coherence of behavioral and physiological response systems, the same was not true of the intensity of sadness. 5
Affect dynamics can also help determine which factors influence emotion duration. For example, Verduyn et al. (2009) aimed to assess two possible predictors of emotional duration: the importance of the emotion-eliciting situation and the intensity of the emotion onset. In the first study, participants were given a daily questionnaire over a 2-week period, and were asked how often they had an episode of fear, anger, and joy that day (0, 1, 2, 3, 4, 5, > 5 times), and using an 8-point Likert scale, to rate the importance of the emotion-eliciting event (0 = not important at all to 7 = very important) and the intensity of the emotion at the beginning of the emotional episode (0 = not intense at all to 7 = very intense). Participants were also asked how long each of these episodes lasted. Here they used a bar with eight intervals, 15 mins each, with a total length of 2 hr to measure emotion duration. It was found that the “more important eliciting situations and higher emotion intensity at onset were associated with longer emotional episodes for all three emotions” (Verduyn et al., 2009, p. 86). A second study with a slightly different methodology (e.g., different emotions with a total length of 1 hr instead of 2 hr) confirmed these results (Verduyn et al., 2009).
It is worth noting that subsequent studies in affect dynamics have also replicated and extended these findings. For example, Verduyn and Lavrijsen (2015) investigated 27 emotions to determine the effects of emotion regulation (e.g., rumination and reflection) and appraisal dimension (e.g., event importance) on emotion duration. In brief, participants were provided a questionnaire with two conceptual clarifications. First, to ensure participants were reporting emotions and not moods, they were told than “an emotion is always elicited by a certain external or internal event, and thus has a clear onset” (Verduyn & Lavrijsen, 2015, p. 121). They were also provided definitions of emotion duration (more on this soon). Each questionnaire involved nine of the 27 emotions, and participants were asked to recall a recent episode of each emotion and to briefly describe the emotion-eliciting event and when it occurred (1 = days ago, 2 = weeks ago, 3 = months ago, or 4 = years ago). They were then asked to rate the emotion duration (days, hours, minutes and/or seconds) and emotion intensity (0 = not intense at all to 6 = very intense). In addition, participants were asked to report the regulation strategies adopted, as well as to rate the emotion-eliciting event on a number of appraisal dimensions (e.g., importance) on a 7-point Likert (from 0 to 6). It was found that rumination and reflection, as well as event importance, correlated significantly with emotion duration.
This is not to say that affect dynamics is without its share of conceptual and methodological restrictions. Most research in affect dynamics conceptualizes the end of an emotional episode as its first return to baseline (Verduyn et al., 2009, 2011, 2012). For example, Verduyn et al. (2009, p. 84) defined emotion duration as “the period between [the] onset point and the moment the emotional experience is no longer felt.” By contrast, the end can also be conceptualized as the moment when the episode is “closed” in the sense that it reaches the baseline in a permanent manner (Frijda et al., 1991; Verudyn & Lavrijsen, 2015; Verduyn, 2021). 6 In their study, Verduyn and Lavrijsen (2015) provided participants with a definition of the end of an emotional episode. Half of the participants were told, in line with the previous study, that “an emotional episode ends as soon as the emotion is no longer felt,” whereas the second half were told that “an emotion ends as soon as one fully recovered from the event” (Verduyn & Lavrijsen, 2015, p. 121). Of these different conceptions of emotion duration, the first is easier to measure in experimental conditions, but has the disadvantage that it fails to account for the way an emotion associated with a particular event may reoccur after initially returning to the baseline. As Van Mechelen et al. (2013, p. 175) note, “One may wonder, however, whether such measures of closure (which are being used in the majority of studies of emotion duration) do not capture (possibly temporary) emotion relief rather than full-blown emotional recovery.” The second, by contrast better captures the length of an emotional episode in response to a particular event, but is significantly harder to measure in practice. Experience sampling, for example, can measure how a subject is feeling at any given moment in time, but it can’t always measure physiological and expressive markers of emotion, such as heart-rate and facial muscle changes.
Such limitations aside, research in affect dynamics captures important aspects of emotional duration and change, and thereby makes significant strides in addressing the broader philosophical critique. The focus of the research is still emotional episodes, but they are no longer seen to last just seconds or minutes. Work being carried out in the field suggests that some emotional episodes can last hours, even days (Brose et al., 2014; Lazarus et al., 2021; Verduyn et al., 2009, 2013). This is a considerable step toward accommodating long-term emotions.
The Temporal Components and Dynamics of Long-Term Emotions
Do affect dynamics have the theoretical and methodological resources to address the philosophical critique? In other words, can it help capture important aspects of long-term emotions that (some) philosophers claim is missing from empirical emotion research?
As it stands, there are two features of long-term emotions that are not captured by research in affect dynamics: their temporal components and dynamics. This is best brought out by focusing on one variant of long-term emotions, that is, sentiments. Following Frijda et al. (1991), we can make a distinction between three types of emotional phenomena: emotions, emotional episodes, and sentiments. Emotions are activated or deactivated states of feelings toward a specific object. They include emotions like anger, sadness, happiness, etc. Emotional episodes, by contrast, are states of emotional involvement with a specific object. Crucially, an emotional episode can contain multiple emotions. For example, the emotional episode you undergo after receiving some bad news may involve fear, anxiety, and anger. Finally, sentiments are understood as dispositions to respond emotionally to specific objects or events. The standard examples of sentiments include love, hate, and grief. One important feature of sentiments to recognize is that they are multitrack dispositions: dispositions to manifest a range of emotional responses depending on the context. For example, as Deonna and Teroni (2012, p. 8) note, “If Juliette loves Romeo, she is not just disposed to feel some specific emotion toward him (erotic ecstasy? warm affection? fawning admiration?), but also grief if things turn sour, pride at his accomplishments, or jealousy at the sight of a rival.”
In what follows, I will argue that research in affect dynamics misses certain temporal dimensions of sentiments by way of demonstrating that the field doesn’t adequately account for long-term emotions. Since sentiments are central to this argument, let me address some prima facie concerns with using sentiments as an example of long-term emotions.
First, one might question whether sentiments, though long-term phenomena, should really count as emotions. The worry here for our proposal is that focusing on sentiments, then, won’t tell us about what is missing from a science of long-term emotions. One reason we might be reluctant to treat sentiments as emotions is that they concern multiple affective components. For instance, it might be that sentiments, owing to their multiplicity, operate at a different (say higher) level to emotions. I appreciate the worry, but I don’t think this is a good enough reason to stop treating sentiments as a form of emotion. The obvious point to note is that paradigm examples of sentiments, for example, love and grief, are treated as emotions on the vernacular category, as well as by (some) philosophers and psychologists. For example, both Goldie (2012) and Ratcliffe (2023) are explicit that grief involves multiple affective components, yet treat it as a long-term emotion. Likewise, both Fredrickson (2013) and Sternberg (2018) take love to involve multiple affective components, but don’t depart from the tradition of treating it as an emotion. To this it is worth adding that emotional episodes also involves multiple emotions, and no one is suggesting this means they should be treated as a completely separate phenomenon from emotions.
Second, and relatedly, one might question whether different emotional episodes (often comprising of distinct affective components themselves) should count as instances of one and the same sentiment. For example, if Juliette feels affection at t1, jealousy at t2, grief at t3, and so on, how can we say that they all belong to the same sentiment, that is, her love for Romeo? The worry here for us is that we might be falsely assuming that her love for Romeo is a long-term emotion by mistakenly classifying emotional episodes that span different time periods as all being instances of her love for him when some simply don’t belong to that sentiment.
Figuring out the identity conditions of sentiments, like those of other affective phenomena, is a tricky issue, which is best addressed another time. However, for now, we can note that emotional episodes are typically individuated based on their intentional profile, roughly, what they are about. For example, we can say that this particular episode of sadness is about a sad film, whereas this other sadness episode is about your recently failed exam. We can, to my mind, demarcate sentiments in an analogous manner. For example, Juliette's love for Romeo manifests itself as affection at t1, jealousy at t2, grief at t3, and so on, but these are all part of the same sentiment because they are all about Romeo. By contrast, her love for her father might also manifest itself in a range of similar emotions, but these are all part of a different sentiment, her love for father, as opposed to say Romeo, because they are about her father. I don’t take this to be the complete story about how to individuate sentiments, but we see that one way to do so is to appeal to their intentional features. 7
Finally, since both emotional episodes and sentiments contain multiple affective components, one might also worry about how to differentiate sentiments from emotional episodes. The precise worry for the present proposal is that, if they are one and the same, then since affective sciences has a long history of focusing on various dimensions of emotional episodes, a case can’t be easily made that they miss the temporal dimensions of sentiments. As I understand it, an emotional episode may contain a multiplicity of emotions, while a sentiment should really be seen to involve multiple emotional episodes themselves. So while a given emotional episode that Juliette undergoes to some event concerning Romeo may involve jealousy, anger and hurt, her overall attitude of love towards him involves the full gamut of emotional episodes she feels towards him. In this way, while two emotional episodes, for example, the jealousy she feels about Romeo at t1 and t20, might not count as the very same emotional episode, say because they are discontinuous and too far apart in time, such episodes will still belong to the wider phenomenon of her love toward Romeo.
There is also an easier, more common, way to draw the distinction between emotional episodes and sentiments. That is, we can treat emotional episodes as emotional responses and sentiments as dispositions to have such responses. For our purposes, however, we can remain neutral as to whether sentiments themselves are patterns of responses or dispositions to manifest such patterns. The important point to stress for now is that long-term emotions, such as sentiments, can be seen to involve a range of emotional episodes, comprising a multiplicity of individual emotions. It is my view that affect dynamics ignores this point, which is precisely why it fails to capture the temporal components and dynamics of long-term emotions. Let us begin by focusing on their temporal components.
Recall the two ways emotion duration is conceptualized in affect dynamics, that is, as either a temporary or permanent return to the baseline. Suppose we use the more liberal conception of emotion duration we have at present, and treat the duration of grief as the length of time it takes for the relevant emotional episode associated with grief to reach its baseline in a permanent manner. While some researchers in affect dynamics acknowledge that a single emotional episode can involve multiple emotions (e.g., Van Mechelen et al., 2013), most of the experimental work carried out in the area does not explicitly recognize this fact and proceed as if individual emotional episodes correspond to single emotions.
For example, Verduyn et al. (2009, p. 84) describe the goal of their first study as investigating “variability in the duration of emotional episodes of fear, anger, and joy, and to account for this variability.” They make no mention of how emotional episodes can contain a multiplicity of different emotions. Moreover, the daily questionnaire they used was divided into three blocks, one for “anger episodes,” one for fear and another for joy. Participants who got the anger block, for instance, had to report how often they experienced anger that day, as well as to rate the intensity of such experiences. Finally, in their results, they report that short episodes of fear where more common than short episodes of anger. All of this gives the impression that they are conceiving of emotional episodes in a way that does not factor in the multiplicity of emotions that can be involved in any given episode. The same could be said for other studies in affect dynamics (Brans & Verduyn, 2014; Verduyn et al., 2012; Verduyn & Lavrijsen, 2015). For instance, Verduyn and Lavrijsen (2015, p. 121) explicitly define emotion duration in terms of the end of “an emotion” (i.e., instead of emotional episodes) in the questionnaires provided to their participants. They also focus on 27 discrete emotion categories, without any acknowledgment that such emotions can come in pairs, tuples, and so on, to form any given emotional episode. If we take our cues from such studies, we would take the end of grief to be the end of an episode of sadness, say to the loss of a loved one. The problem is that if we insist on investigating grief in this way, we miss the range of emotions that might be involved in long-term grief, such as frustration, anger, anxiety, and so on. In other words, we miss the temporal components of that sentiment.
Recognizing the complex nature of emotional episodes takes us some way toward addressing this concern, however, it won’t be resolved completely given the conception of emotion duration employed in affect dynamics. Say we recognize that an episode of grief can involve anger, fear, anxiety, as well as sadness. Studying an episode of grief in this way can tell us about the temporal components of that particular grief episode, but it won’t tell us about all the possible components involved in grief. For example, it might be that episodes of grief experienced at the initial stages of bereavement involve surprise, fear, and anxiety, whereas subsequent stages involve a range of permutations, for example, sadness and anger, sadness and guilt, relief, and so on. Whether or not this is the case is not what is important here. Rather, the point is that this is not something we can discern if we construe the end of grief as the end of a handful of grief episodes. Speaking generally, affect dynamics conceives of emotion duration in such a way that we might still fail to recognize the full diversity of emotional components involved in a given sentiment.
Now let us consider the repercussions of this conception for determining the temporal dynamics of long-term emotions. If sentiments involve a range of emotional episodes, which in turn comprise a range of individual emotions, the aforementioned way of conceptualizing emotion duration will also fail to uncover the various ways the emotions involved in sentiment change over time. To clarify, the issue in this particular case is not (just) that we might fail to accommodate the full diversity of emotional components of a given sentiment, but rather that we might also fail to uncover the various patterns in the way such emotional components emerge and dissipate over time. For example, suppose we investigate the temporal dynamics of hatred by measuring how the relevant emotion associated with this sentiment, say anger, increases or decreases over time. This factors in one aspect of the temporal dynamics of hatred, but neglects others. The anger involved in your hatred toward your stepfather might, for instance, fade over time, whereas your sense of resentment may increase. This is not something we can account for by simply focusing on the temporal dynamics of anger.
Once again, recognizing that emotional episodes can involve a range of distinct emotions takes us some way toward addressing this concern, but it won’t address it completely. If we study episodes of hatred and allow for such episodes to compose of a range of emotions, we might uncover that both anger and resentment are involved in hatred, and perhaps even that one of these emotions decreases while the other increases as the relevant episodes unfold over time. However, by conceiving of the end of hatred as the end of a handful of emotional episodes, we miss the complete picture of how the various emotions involved in hatred unfold over time. In other words, we miss their temporal dynamics. For example, hatred might comprise three core emotions, anger, disgust and resentment, and it might be that one of these emotions, say disgust, appears mainly at the early stages of hatred and gradually decreases over time, whereas another emotion, say resentment, only tends to appear at the later stages, and gradually increases from there. Of course, this is all speculative, but the point is that these sorts of temporal dynamics aren’t things we can discern if we conceive of the end of hatred as the end of a few episodes of hatred. All in all, conceptualizing the end of a sentiment according to even the most liberal conceptions of emotion duration in affect dynamics means we miss out on certain temporal dynamics concerning that sentiment.
Where does this leave us with respect to affect dynamics? I think we can acknowledge that affect dynamics is an improvement on previous emotion research in the sense that it helps capture the temporal components and dynamics of emotional episodes (e.g., anger, fear, and sadness). However, we should also recognize that research in this field falls short when it comes to investigating the temporal components and dynamics of sentiments (e.g., love, hate, and grief). Thus far, affect dynamics does not investigate such sentiments. What's more, it appears that the theoretical tools it employs, in particular, the way it presently conceives emotion duration, do not lend themselves to studying the temporal components and dynamics of such phenomena.
It is worth stressing that raising this worry does not rest on adhering to any specific philosophical theory of emotion. Rather, it is a direct fallout of the way emotion duration is conceived of in affect dynamics. In brief, it equates emotion during with the duration of emotion episodes, and thereby fails to address the duration of emotional phenomena that may last longer (e.g., sentiments). Put simply, affect dynamics does not yet recognize just how long some emotional phenomena last. What I have argued is that this failure results in our missing the temporal components and dynamics of such phenomena.
Toward an Integrative Approach to Long-Term Emotions
How might affect dynamics address this worry? An obvious answer is to reconceptualize the way it defines emotion duration. Psychologists working in affect dynamics are encouraged to adopt an even longer view, and take seriously the assertion made by (some) philosophers that emotions, at least some of them, last months, if not years.
Crucially, this involves going beyond the old dichotomy, which treats emotions as either brief episodes, or longstanding dispositions to manifest such brief episodes. There is a temptation, both in some areas of philosophy, as well as in psychology, to treat long-term emotions as simply dispositions to manifest short-term emotions. As we have seen, this obscures the complexity of long-term emotions. For example, it makes us miss both the temporal components and temporal dynamics of long-term emotions.
Philosophers who recognize this worry often propose remedies that involve reconceptualizing emotion itself. For example, for Merleau-Ponty (1964, p. 53), emotions are “a variation in our relations with others and the world.” This is certainly one to ensure that we don’t treat all emotional phenomena as short-lived emotional episodes. However, such conceptualizations, as found in the phenomenological tradition, can’t be easily operationalized, and thereby are unlikely to help with empirical inquiry.
A less demanding remedy is to simply recognize (i) that some emotional phenomena, say those involved in bereavement, may last more than just days, weeks, or even a few months, and (ii) with this also keep in mind that such phenomena involve multitrack dispositions that can result in a myriad of different emotional responses. 8
There are a few different ways to reconceptualize emotion duration to satisfy these conditions. First recall the second more inclusive way emotion duration is conceptualized in affect dynamics: [T]he end of an emotional episode can be equated with the moment in time at which the episode is ‘closed’ in the sense that the emotion intensity associated with an emotion-eliciting event reaches zero (or baseline) in a permanent manner. (Verduyn, 2021, p. 13–15) The end of a sentiment can be equated with the moment in time at which all the relevant episodes are ‘closed’ in the sense that the emotion intensity associated with the emotion-eliciting event reaches zero (or baseline) in a permanent manner.
On this way of construing the duration of sentiments, a sentiment like love ends when all the emotional episodes associated with it (e.g., fear that your beloved might come to harm, sadness that they are suffering, and so on) come to an end in a permanent manner.
Let me make some clarificatory points before explaining how we might apply this definition in practice. First, I doubt there is a fixed moment in time when a sentiment can be said to end, for example, when a subject stops grieving or falls out of love. Grief may never end. Moreover, there might not be a clear temporal interval when one starts and stops to manifest all the emotional episodes associated with love. The boundaries will, in all likelihood, be vague. Nevertheless, the new definition provides an operationalizable conception of the duration of sentiment that can aid future research in affect dynamics. What's more, it gives us a definition that can be used alongside the definition of emotional episode duration already on offer in this field.
Second, we will need a way to tell when an emotional episode is an instance of a particular kind of sentiment and when it isn’t. However, this isn’t any different from the challenge of demarcating emotions in general. As noted earlier, emotions tend to be individuated by their intentional profile, that is, what they are about (see pp. 6 and 13). Emotion research in affect dynamics typically employs broad, vernacular definitions, which aligns with this way of individuating emotions, for example, sadness is treated as a response to loss, say of a child (Verduyn, 2021). We can appeal to analogous definitions in the case of sentiment, with crucial caveat that we allow sentiments to comprise a range of emotion types. For example, instead of treating grief as sadness felt in response to the loss of a loved one, we can have a more liberal conception that treats grief as any emotions we have in response to the loss of a loved one. On this way of conceptualizing grief, we can classify emotions like anger and anxiety as episodes of grief, that is, insofar as they are responses to such a loss. Precisely how to define specific sentiments will prove contentious, but this is not our main concern here. 9 What we want for now is some plausible way of defining the sentiments of interest, which can be used alongside our operationalizable conception of sentiment duration.
One final thing that remains unclear worth mentioning is how the proposed account of the duration of sentiment fits in with a dispositional account of emotion. Thus far, I have tried to remain neutral on the metaphysically contentious issue over whether emotions are states or processes. However, if a sentiment is supposed to end when all the emotional episodes associated with it are closed, this seems to run counter to a dispositional account of sentiment, as dispositions can, in principle, persist without ever being manifest. The first thing to note in response is that the proposal provides a working definition that is supposed to aid psychological experimentation, as opposed to being a metaphysically substantive thesis about when sentiments end tout court. A sentiment like grief presumably ends on a dispositional account when one is no longer disposed to manifest any of the emotional responses associated with grief. One might grant this metaphysical thesis and also accept that a practical way to expand research in affect dynamics such that it can also focus on sentiments is to supplement existing definitions of the duration of emotional episodes with our working definition of the duration of sentiment. There need not be any contradiction in achieving these differing aims. Second, there may also be ways of accommodating a dispositional account of sentiments more explicitly in a working definition of their duration:
The end of a sentiment can be equated with the moment in time at which a subject is no longer disposed to manifesting any of the relevant emotional episodes in the relevant contexts. The end of the emotional episodes associated with that sentiment can be equated with the moment when all the relevant episodes are “closed” in the sense that the emotion intensity associated with the emotion-eliciting event reaches zero (or baseline) in a permanent manner.
A definition such as this might enable us to grant the dispositional nature of sentiments while also spelling out an operationalizable way to track when the relevant episodes associate with a sentiment are closed. The thought is that we recognize, re-(a), that sentiments might not end even when all the relevant emotional episodes stop being manifest, but psychologists studying sentiments can, re-(b), take as their immediate aim capturing the temporal dynamics involved in such episodes, that is, until these episodes reach their end. Working out the precise details can be left for a later date. For now, what remains clear is that there are ways to accommodate sentiments into research in affect dynamics regardless of our background metaphysics.
Now let us turn to how we might actually employ our working definition (either one) to satisfy the relevant conditions. In practice, satisfying conditions (i) and (ii) means supplementing existing research on affect dynamics with longitudinal data, which involves emotional patterns that last not just minutes, hours or days, but weeks, months, even years. Crucially, such a science must also employ comparative analyses to determine how such processes vary over these longer time spans. The good news is that once we remove the conceptual obstacle created by existing definitions of emotion duration, there are no further barriers to a science of sentiment.
For example, recall the experiments carried by Verduyn et al. (2009) which aimed to assess the relationship between emotion duration on the one hand, and the importance of the emotion-eliciting situation and the intensity of the emotion onset on the other. In the first study, participants were given a daily questionnaire over a 2-week period and they were then asked how often they had an episode of fear, anger, and joy that day (0, 1, 2, 3, 4, 5, > 5 times), to rate the important of the emotion-eliciting event (0 = not important at all to 7 = very important) and the intensity of the emotion at the beginning of the emotional episode (0 = not intense at all to 7 = very intense), and how long each of these emotions lasted. In terms of the last question, they used a bar with eight intervals, 15 mins each, with a total length of 2 hr to measure emotion duration. They found a positive correlation between emotion duration and the two predictors.
A study that better captures the temporal components and dynamics of sentiments can keep the broad experimental framework of the original study but make the following revisions. Participants can be given a daily questionnaire over a 2-week period, but for 4 months of the year (e.g., January, April, July, and October), over the course of 3 years. Participants can also be provided with explicit definitions of the sentiments in question. For example, they can be told that “grief is an emotional response to the loss of a loved one, and can include feelings of sadness, anger, fear, and anxiety.” Participants can then be asked how often they had the emotional episodes involved in that particular sentiment. For example, they can be asked, “how often did you feel anxious in response to the loss of your loved one (0, 1, 2, 3, 4, 5, > 5 times)?” The question can be repeated for the other relevant emotions, that is, sadness, fear, anger, and so on. Moreover, since participants are provided with a daily questionnaire, the length of an emotional episode can be expanded to 24 hr instead of 2 hr.
This type of study design is based on our working definition of sentiment duration, with a view to satisfying conditions (i) and (ii). Since the proposed study lasts 3 years, re-(i), it takes into account that some emotional phenomena, such as grief, can last more than just a few days, weeks, or months. Moreover, since sentiments like grief are explicitly defined in a way that draws attention to their various emotional components, re-(ii), it also grants that sentiments involve multitrack dispositions that can result in a myriad of different emotional responses. It is worth noting that these conditions are met without compromising the overall aim of the original study: to explain the variability of emotion duration. As before, we can assess the impact of two possible predictors on emotion duration. However, we do so with the added benefit of capturing the relationship between these predictors and long-term emotional phenomena like sentiments.
Of course, longitudinal studies are difficult to conduct in practice because they are costly, involve significant time-commitments on the part of participants, and suffer from attrition bias. But the good news is that the sort of longitudinal studies that can inform a science of sentiment is not without precedence. In fact, they are already a few existing experiments that speak to conditions (i) and (ii). For example, Wrzus et al. (2014) aim to study the affect dynamics across the lifespan by measuring one physiological marker of affect, namely heart-rate, across participants aged 14 to 83. They found that heart-rate reactivity decreased with age, whereas hate-rate recovery time increased. Ideally, similar studies will measure a range of emotional markers (e.g., phenomenological, physiological, expressive, and behavioral). Moreover, they will measure intrasubject variability of these markers, as well as intersubject variability. In other words, we want to see not just how subjects of different ages respond emotionally, say to the loss of a loved one, but also how an individual responds emotionally to such conditions over the course of a lifetime. Such research has the potential to satisfy condition (i).
There is also work done in affect dynamics that lends itself to satisfying condition (ii). For example, consider research on between-episode affect dynamics, which aims to capture “how emotions and moods at one point in time may influence subsequent emotions and moods, and the importance of the time-scales on which we assess these dynamics” (Verduyn & Lavrijsen, 2015; Waugh & Kippers, 2021, p. v). This is a step in the right direction. What is missing is the recognition that different emotional episodes, which may or may not influence one another over time, might also belong to a broader sentiment. For example, your emotional episode of sadness today might have a bearing on the episode of anger you have in a couple of days; however, they might also both be instances of a wider sentiment, such as grief.
The overall lesson here is that capturing what philosophers claim is missing from a science of long-term emotions doesn’t involve a radical shift in how researchers in the field of affect dynamics presently study emotion. Instead, what is required are some thoughtful modifications to existing studies while taking certain conceptual clarifications into account. In particular, I argued that a way to remedy present omissions (i.e., a lack of focus on the temporal components and dynamics of sentiments) is to broaden our conception of emotion duration and make use of longitudinal data.
One thing I have not touched upon is how such research might bear on specific philosophical theories of emotion. Will an affective science that measures long-term emotional patterns capture all the ways emotions, say, situate ourselves in the world? To my mind, this is not something we can judge prematurely. Whether we can have a full integration of any given philosophical theory and a psychological science of long-term emotions, or whether some differences will prove irreconcilable, is something we can only discern when such a science reaches maturation.
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
