1 Introduction

Loneliness is on the rise in Western societies (Olds and Schwartz 2009; Twenge 2017; Murthy 2020). This is quite worrisome, since loneliness is typically correlated with a host of negative consequences, including psychological, physical, and societal problems (Cacioppo and Patrick 2009; Olds and Schwartz 2009; Wang et al. 2017; Murthy 2020). Taking note of this, some scholars have recently started exploring the ethical and political issues that loneliness generates (see Brownlee 2016; Valentini 2016; Cormier manuscript). In this paper, I want to address a different and more foundational question: What is loneliness?

It is commonly agreed in the literature that loneliness is distinct from solitude and from social isolation. The latter are often conceived as objective states. Solitude is the state of being alone, that is, not surrounded by any other individual.Footnote 1 Social isolation is the state of having no social contacts or relations. Loneliness, by contrast, is a subjective state – more specifically, it is a negative affective state. One may feel lonely in this sense even when surrounded by people or within a relationship. Symmetrically, one may be alone or socially isolated without feeling lonely.

This paper is an attempt to elucidate the nature of loneliness thus conceived. I will proceed as follows. In the first part (Sect. 2), I will examine in considerable detail Tom Roberts and Joel Krueger’s recent account of loneliness (Roberts and Krueger 2021). Besides being one of the few philosophical accounts of loneliness that is developed at sufficient length,Footnote 2 it is also, and most importantly, a very perceptive account, one that well captures certain aspects of our ordinary experience of loneliness. I think we can learn a lot from it. That said, in what follows I will raise three important challenges to Roberts and Krueger’s account. This will pave the way for my own account of loneliness, which I will present in the second part of the paper (Sect. 3).

2 Three Challenges for Roberts and Krueger’s Account of Loneliness

2.1 Roberts and Krueger’s Account

When talking about emotions, it is common to distinguish between occurrent emotions and dispositional emotions. Occurrent emotions are mental episodes, which typically have a distinctive phenomenology. Dispositional emotions are dispositional states that have occurrent emotions as their manifestations. The term ‘loneliness’ can refer either to a dispositional emotion or to an occurrent emotion. If dispositional emotions are dispositions to experience occurrent emotions, however, it seems that, to understand what the dispositional emotion of loneliness is, we need to first understand what loneliness is as an occurrent emotion. The latter is also Roberts and Krueger’s (R&K, henceforth) aim (p. 191).Footnote 3

R&K conceive of occurrent loneliness (henceforth, simply ‘loneliness’) as “an emotion that essentially concerns absence” (p. 186). It is not the only emotion of this kind, though. In fact, loneliness belongs to a family of emotions that “take absences as their objects” (p. 187), that is, emotions that “are essentially about what is missing, out of reach, or nonoccurring” (p. 186; italics in the text). Emotions of absence share two elements. In R&K’s own words, “[t]he first element of each state is a pro-attitude towards some absent thing or quality, such as a desire for it or an attitude of admiration, lust, or appreciation. The second element is a complex awareness that the absent thing cannot be made to be present; it cannot easily be achieved, generated, or brought about” (p. 187).

Examples of emotions of absence are homesickness, unrequited love, envy, nostalgia, and social anxiety. What distinguishes loneliness from these emotions is the fact that loneliness is essentially about certain absent social goods (p. 191, p. 199). As examples of social goods involved in episodes of loneliness, R&K mention “companionship, moral support, physical contact and affection, sympathy, trust, romance, friendship, and the opportunity to act and interact – and so to flourish – as a social agent” (p. 191). According to this picture, then, the two elements characteristic of emotions of absence take the following form in loneliness. The first element is a pro-attitude towards one or more of these social goods. The second element is an awareness that these social goods are “missing and out of reach, either temporarily or permanently” (p. 186).

2.2 The Challenge from Descriptive Adequacy

The first challenge that I want to raise is what I call a ‘challenge from descriptive adequacy’. Roughly, the idea is that, in some cases, R&K’s account does not fit our considered judgments about what counts or does not count as an episode of loneliness.

As we have seen, R&K’s main claim is that loneliness has a “two-part emotional structure” (p. 190). There are various ways to understand this claim. I will consider three of them in what follows.

2.2.1 First Interpretation

According to a first interpretation, loneliness just is the combination of a pro-attitude and a cognitive state of the relevant sort. On this reading, the two elements identified by R&K are individually necessary and jointly sufficient for an individual to be in a state of loneliness. The resulting account is an instance of the ‘desire-belief’ theory of emotions (provided that the terms ‘desire’ and ‘belief’ are given a broad meaning). I am not certain that this is the correct interpretation of R&K’s account. In footnote 6, they express a willingness to remain neutral about different theories of emotions. The present interpretation does not square with the neutrality they aim for. Even so, I think it is instructive to examine whether this is a plausible account of loneliness.

One problem is on the sufficiency side. Consider the following example. Suppose that an individual desires that his partner opens up about his past but realizes that this is not going to happen any time soon. This individual appears to undergo the two elements of loneliness. He has a pro-attitude towards an interpersonal good, i.e., intimacy, as well as an awareness that this good is currently missing and might be missing for a while. If a pro-attitude and a cognitive state of this sort are sufficient for loneliness, then we should conclude that the individual feels lonely. But we can imagine cases where this is not true. Suppose, for example, that the relationship is at an early and enthusiastic stage. The individual may tell himself: “It will take some effort, but I like my partner so much and I really want to invest myself in this relationship, so that we can get more intimate with each other”. It seems that the individual in this example does not feel lonely.Footnote 4 He is enthusiastic and motivated by his romantic love, even though he realizes that some social good is presently missing and out of reach. If this is true, it follows that having a pro-attitude towards some social good and an awareness that this good is missing is not sufficient for undergoing an emotion of loneliness.

2.2.2 Second Interpretation

This leads us to a second interpretation of R&K’s account. Perhaps the two elements are only necessary components of loneliness. However, even this interpretation is problematic. In this subsection, I will specifically focus on the claim that a pro-attitude towards some social goods is a necessary component of loneliness. (I will focus on the claim that the individual’s awareness is a necessary component of loneliness in subsection 2.2.3.) As examples of relevant pro-attitudes, R&K mention desires, craves and yearns (p. 198). These are all motivational states. If so, it appears that there are possible counterexamples to R&K’s necessity claim. It seems indeed possible for an individual to feel lonely at a particular time without also having an occurrent motivation to bring about the missing social good. This is especially true in cases of chronic loneliness and depression, when an individual’s motivational resources are drained.

R&K might react to this alleged counterexample in one of the following ways. First, they might say that, in order for an individual to undergo an episode of loneliness, it is not necessary that they have an occurrent desire to have or bring about a social good. Rather, it is necessary that they have a dispositional desire to do so, that is, a disposition that manifests into occurrent desires to have or bring about the relevant social good in some circumstances, though not necessarily at the time the subject experiences loneliness. The problem with this reply is that the pro-attitude now appears to be a pre-condition for experiencing loneliness, rather than a necessary component of an episode of loneliness.

Second, R&K might say that what matters for loneliness is not that the individual is motivated to have or bring about some social good, but that the individual values that social good, where valuing a social good does not necessarily involve a motivation to have or bring about the social good (although it might in some instances). Accordingly, in order for an individual to undergo an episode of loneliness, it is necessary that the individual has an occurrent valuing attitude (rather than an occurrent desire).

The prospects for this move depend on how valuings are conceived. There are various accounts in the literature. Without entering into details, for present purposes it is sufficient to notice that the main alternatives to motivational accounts of valuings hold that valuing an item requires judging that item to be valuable (see, e.g., Scheffler 2010; Dorsey 2012; Tiberius 2018).Footnote 5 If we insert this account of valuings into the interpretation of R&K’s account of loneliness under consideration, then it follows that in order for an individual to undergo an episode of loneliness, it is necessary that the individual judges, at the time of the episode, that the missing social good is valuable.

The problem with this account is that it is subject to potential counterexamples. To begin with, it seems intuitively possible for an individual to experience loneliness without making value judgments. This possibility becomes even clearer if we admit that at least some non-human animals and young children are capable of experiencing loneliness. Non-human animals and young children typically lack the conceptual resources to make value judgments (i.e., they lack evaluative concepts, which are required to make value judgments). So, on this account, they could not experience loneliness. But it seems that they can, e.g., think about your companion dog waiting all day for you to come back from work.

There is a third reply that R&K may give in response to the challenge that a pro-attitude towards some social good is not necessary for loneliness. I think this is their actual reply. They discuss it in the section on ‘Chronic Loneliness’ (pp. 199–201). There, R&K recognize that, in instances of chronic loneliness, an individual may lose all their desires for social goods. This loss leads the individual to undergo what they call an “affective flattening”. They then suggest that “[t]his flattened affective condition […] just is the way in which loneliness manifests itself when the subject has lost all interest in the spectrum of social goods towards which we are usually positively oriented” (p. 201, italics in the text).

If I understand well, this response involves two claims. The first is that, due to the lack of the relevant pro-attitude, the chronically lonely individual will not be able to experience occurrent tokens of loneliness. I take this to follow from the claim that the individual is affectively flattened. The second is that we can nevertheless see the chronically lonely individual as manifesting loneliness through their “diminished feeling, motivation, and attentiveness” (p. 191). Combined, these claims suggest that a pro-attitude is necessary for undergoing feelings of loneliness, but not for manifesting loneliness. If this is the correct reading, then there are two problems with it. The first is that it seems implausible to say that a chronically lonely individual cannot feel lonely. The second is that it is unclear how the chronically lonely individual’s affective flattening can be a manifestation of the dispositional state of loneliness if the individual is incapable of experiencing any episodes of loneliness.

2.2.3 Third Interpretation

If the previous considerations are on the right track, then a pro-attitude towards some social goods is neither a sufficient nor a necessary component of loneliness. This leads us to a third interpretation of R&K’s account. This interpretation holds that, while having a pro-attitude towards some social good may be a pre-condition for experiencing loneliness, the episodes of loneliness themselves only involve an awareness that the relevant social good is missing. I do not think this is what R&K have in mind, but it is instructive to examine this interpretation, as this helps getting closer to the account of loneliness that I believe to be correct.

One way to formulate this third interpretation is by saying that the awareness that a social good is missing is the sole component of an episode of loneliness, i.e., this is what an episode of loneliness consists in. However, this claim can be challenged. To see why, note that there are two ways in which the individual’s awareness can be conceived. One is as a non-evaluative cognitive state, that is, as a state that represents an item (i.e., a social good) as possessing some non-evaluative property (i.e., the property of being missing or absent). The second is as an evaluative cognitive state, that is, as a state that represents an item (i.e., an absent social good) as possessing some evaluative property (i.e., the property of being bad in some specific way). I think that R&K conceive of the awareness that is involved in episodes of loneliness in the former sense. If so, it is possible to raise the following objection. It is often claimed that emotions are kinds of evaluations, in the sense that they are either evaluative attitudes or mental states with an evaluative content. Yet, if loneliness consists in a non-evaluative kind of awareness, it follows that loneliness is not an evaluation in either of these senses. Therefore, if emotions are indeed evaluative states (as I think they are), this interpretation of R&K’s account should be rejected.Footnote 6

An alternative way to formulate the third interpretation of R&K’s account consists in saying that the awareness that a social good is missing is a necessary component of an episode of loneliness, though not the sole component. The other components bear the burden of accounting for the evaluative character of loneliness. The problem is that, since no motivational state is necessary for loneliness, it appears that loneliness’ evaluative character must then be explained by appealing to some evaluative cognitive state, e.g., an evaluative perception or an evaluative attitude, concerning the missing social good. But if this is the case, what role is there left to play for the awareness that a social good is missing? The most plausible answer – it seems to me – is that such an awareness is one of the cognitive bases of an episode of loneliness. That is, it is one of the states by means of which loneliness apprehends its object. Arguably, however, the cognitive bases of a given emotion are elements that, though necessary to undergo the emotion, do not constitute the emotion. If this is true, then the individual’s awareness that a social good is missing is not what loneliness is made of, even though it may be what an episode of loneliness is necessarily caused by – at least provided that the term ‘awareness’ is given a disjunctive meaning, such that to be aware that a social good is missing is to either believe, or perceive, or imagine, or suppose, etc., that a social good is missing.

2.3 The Challenge from Phenomenal Intensity

It is a feature of our ordinary experience that loneliness admits of degrees. By this, I mean that loneliness can be felt with different intensities. To be complete, an account of loneliness must explain what determines the phenomenal intensity of different episodes of loneliness. In this subsection, I will raise some worries about R&K’s proposed explanation. I call this ‘the challenge from phenomenal intensity’.

R&K’s account of phenomenal intensity is first presented in Sect. 2 of their paper, where they illustrate the main characteristics of the emotions of absence. R&K write: “[t]he intensity of the emotion is, in each case, determined by how strongly the agent cares about the missing good” (p. 189). This account also applies to loneliness, as an emotion of absence. Accordingly, the intensity of an episode of loneliness is determined by the strength of the pro-attitude towards the social good. Let us leave the issue of whether the pro-attitude is really a necessary component of loneliness aside. Suppose for the sake of argument that it is. The question is whether this account is plausible in its own terms.

One reason to challenge this account is quite general. Loneliness is a paradigmatically unpleasant emotion. R&K offer an account of its unpleasantness that belongs to the family of desire-theories of unpleasantness. It may be argued, however, that desire-theories of unpleasantness are not especially plausible (see, e.g., Bain 2013). When they are conceived as purely motivational states, desires (and similar pro-attitudes) do not seem to have the kind of phenomenology that emotions have. For example, I now have a desire to finish writing this paper. This desire has certainly a particular motivational force. However, it does not have a distinctive phenomenology like emotions do. Things are different if we consider urges or cravings, instead of desires. It seems to me that urges and cravings do have a distinctive phenomenology. Yet, I think it is implausible to reduce the class of pro-attitudes that are supposed to constitute loneliness to urges and cravings.

There is an additional, more specific reason to challenge R&K’s account, namely, that their account seems subject to counterexamples. Let us examine again the case of the individual who desires that his partner open up about his past. Suppose that this desire is especially strong. Suppose also that, after several efforts, the individual’s partner opens up quite a bit, in comparison to the beginning of the relationship, though not as much as the individual would want. In this case, R&K’s account implies that the individual feels very lonely, on the ground that he cares a lot about a good (i.e., intimacy) that is not fully realized. It seems to me, however, that it is possible to imagine circumstances in which this is not true. The individual might feel quite proud about his partner for making all these efforts. Or he might feel happy that the relationship is much more intimate than when it started. In fact, even assuming that the individual feels lonely in relation to that aspect of their relationship, his loneliness might not be especially intense. For example, this may be true if the relationship is otherwise very satisfying.

One feature of the previous example is that the individual has not just a strong pro-attitude towards some social good and an awareness that the social good is missing, but also an awareness that the realization of the social good is not that far ahead. This remark suggests a possible amendment of the previous account. One possibility is that loneliness’ phenomenal intensity is jointly determined by the strength of the pro-attitude towards some social good and by the degree of unattainability that the social good is perceived to have. I think that there is textual evidence in favor of this modified account. Indeed, at some point, R&K claim that “[episodes of loneliness] are experiences whose subjective unpleasantness is determined by how painful it is to crave certain basic human needs and see no hope of them being fulfilled” (p. 198, my italics). It seems to me that this statement is too strong. It is not necessary for loneliness that the individual be completely devoid of hope about the attainability of the missing social good. Nor that hopelessness be involved.Footnote 7 That said, it certainly seems plausible, at first sight, that how lonely an individual is partly depends on how difficult the individual perceives the attainment of the missing social good to be.

However, I think that some potential counterexamples cast doubt even on this revised account. Let us go back to our romantically involved individual. Let us suppose that he strongly desires to have some relationship good, e.g., a deep sense of trust with his partner. This time, however, let us suppose that, knowing his partner’s traumatic background, he is very aware that they might never be able to reach the kind of mutual trust that he desires them to have. The relationship being at the usual early and enthusiastic stage, he nevertheless fully commits to it and invests time and energy in trying to accompany his partner during the difficult process of becoming able to trust someone again. R&K’s account implies that the individual feels very lonely. But it seems to me that, in the described circumstances, this may not be the case. The individual has very realistic expectations, but these do not translate into loneliness, or into an intense experience of loneliness. If this is true, the upshot is that the strength of the pro-attitude and the perceived unattainability of the missing social good is not what determines the phenomenal intensity of loneliness.

2.4 The Challenge from Demarcation

The last challenge I want to raise to R&K’s account of loneliness is a ‘challenge from demarcation’. In a nutshell, the challenge is that R&K’s account has trouble demarcating loneliness from other negative emotions.

At the beginning of their paper, R&K claim that what distinguishes different emotions from each other, and from other affective states, is their intentional content (p. 186). They later claim that what distinguishes loneliness from other emotions of absence is the fact that loneliness concerns certain specific social goods. Combining these remarks, we can say that what distinguishes loneliness from other emotions is its intentional object, where the intentional object is the object to which emotions are directed. Instances of loneliness always have a missing social good as their intentional object in this sense.

I think there are three problems with this claim. The first is that the social goods that R&K mention in their paper are quite heterogeneous. It is unclear what unifies them. They are, of course, ‘social’ goods. But the category of social goods also includes goods that are not typically involved in instances of loneliness. For example, it may include the collective pleasure of winning the final of the World Cup, a social good whose perceived absence does not seem to generate loneliness, though it certainly generates disappointment. The first problem is, thus, that it is unclear what the different social goods with which loneliness is concerned have in common, in virtue of which loneliness can be distinguished from other emotion types.Footnote 8

The second, related problem is that the social goods which loneliness is typically about are sometimes the objects of other negative emotions that one may experience within romantic, friendship or social relationships. For example, lack of trust, intimacy, or familiarity may cause, and be the intentional objects of, emotions such as sadness and disappointment vis-à-vis one’s relationships, which are relevantly distinct from loneliness. To put this differently, an individual who desires intimacy and realizes that intimacy is unattainable may experience disappointment about the lack of intimacy in their relationship, rather than loneliness.

R&K might reply that what demarcates loneliness from disappointment as distinct emotion types is the fact that loneliness is essentially concerned with social goods, whereas disappointment is not. One may indeed be disappointed about failing an exam. While this is a missing good, it is not a missing social good. One implication of this reply is that the intentional content of a token of loneliness is not sufficient to distinguish it from a token of a different emotion type. For a token of loneliness may have the same intentional content as the token of a different emotion type. In order for an emotion token to count as an instance of loneliness, an additional condition must be satisfied: on top of having a missing social good as its object, it must also be the case that that token could not have had an intentional object of a different kind, i.e., a missing good of a different variety.

I do not think this is a problem for R&K’s account per se – though it is worth noticing that there are accounts that allow us to distinguish the emotion type to which a token belongs purely on the basis of its content, as we will see below. However, this account becomes problematic if there exist other types of emotions of absence that are essentially about missing social goods. For if this is the case, then R&K’s account does not have the resources to adequately distinguish tokens of loneliness from tokens of these other emotion types. R&K themselves offer one such example: social anxiety. Roughly speaking, we can say that social anxiety involves a kind of fear of being judged or disliked by other people. Being positively judged, appreciated, or simply recognized by other people is a social good in the sense that R&K have in mind. Importantly, its (perceived, believed, imagined, or supposed) absence seems essential to social anxiety. It follows that social anxiety is essentially about some missing social good. It is clear, however, that an individual may feel socially anxious without feeling lonely. But if this is true, then R&K do not have the resources to distinguish social anxiety from loneliness.

3 Towards a Receptive Account of Loneliness

The previous discussion highlights a few desiderata for a plausible account of loneliness. The latter must explain (a) what the intentional object of loneliness is, (b) what, if any, is the role of pro-attitudes and other cognitive states, (c) how we can distinguish loneliness from other emotion types, (d) how we can account for the phenomenology of loneliness and its coming in degrees, and (e) how we can account for chronic loneliness. In this section, I begin to sketch my own account of loneliness and show how it can meet these desiderata.

3.1 The Receptive Theory of Emotions

As we have seen, R&K appear to endorse a desire-belief account of loneliness. In the previous section, I offered some reasons to think that this is an inadequate account of loneliness. But there are reasons to think that the desire-belief approach is also inadequate for the purpose of characterizing other emotions. These reasons are well-documented, so I will not rehearse them here (see Deonna and Teroni 2012). Instead, I will briefly present the approach that I consider most promising, namely, the receptive theory of emotions that has been recently put forward by Christine Tappolet (2022).

The receptive theory is a direct descendant of the perceptual theory of emotions. In fact, Tappolet was one of the main proponents of the latter theory. According to her version of the perceptual theory (most thoroughly stated in Tappolet 2016), emotions are perceptual experiences that non-conceptually represent certain objects as possessing specific evaluative properties. Thus, for example, an instance of admiration towards a musician consists in a perceptual experience that non-conceptually represents that musician as admirable. An instance of fear of a dog consists in a perceptual experience that non-conceptually represents that dog as fearsome. And so on.

Over the years, the perceptual theory has been subject to some important objections (see, e.g., Salmela 2011; Deonna and Teroni 2012; Brady 2013). The critics’ main strategy has been to point to various disanalogies between emotions and sensory perceptions. One disanalogy is that emotions have cognitive bases. As we have seen above, this means that emotions require other cognitive states, such as beliefs, sensory perceptions, and suppositions, for apprehending their intentional object. For example, a typical episode of fear of a dog is preceded, and caused, by a visual experience of a dog or by a belief that a dog is present. None of this is true of sensory experiences. A second disanalogy is that emotions have normative reasons. It makes sense to ask an individual who is afraid of a dog the reasons why they are afraid. The demand here is for a justificatory explanation, not a merely causal explanation. By contrast, demanding the reasons why an individual has some perceptual experience does not seem to make much sense. True, perceptual experiences may be veridical or non-veridical – arguably in the same way as emotion tokens can be correct or incorrect. But they are not the kind of mental states for which it makes sense to ask for a justification.

The traditional replies to these challenges have been to advocate for a more ‘liberal’ understanding of perceptual experiences or to characterize emotions as ‘quasi-perceptions’ (see Tappolet 2016). Both these moves, however, have been criticized as being ad hoc. What is a quasi-perception after all? And what independent reasons do we have to liberalize our understanding of perceptual experiences?

The receptive theory emerges in this context as a refinement of the perceptual theory. The receptive theory preserves three core claims underlying the perceptual theory. The first is that emotions provide the subject experiencing them with access to value, in the sense that they reveal or disclose to the subject the evaluative properties that the objects of emotions possess – genuinely, if the emotions that the subject experiences are correct; only apparently, if they are incorrect. The second claim is that emotions do that by representing their objects as possessing these evaluative properties. In other words, emotions have evaluative content. The third is that emotions represent evaluative properties non-conceptually. This explains how even subjects who lack evaluative concepts, such as infants and non-human animals, can experience emotions.

The crucial claim underlying the receptive theory is that emotions provide only mediated or indirect access to value. They disclose value through the mediation of the cognitive states necessary for apprehending their intentional objects. Since this is a feature that distinguishes emotions from standard perceptual states, the receptive theory agrees that emotions are not perceptual states themselves. It only claims that they are like perceptual experiences in terms of the epistemological role they play with respect to value and in terms of their having non-conceptual content. But they are unlike perceptual experiences in that they are not at the sensory periphery, and they do not provide unmediated access to the properties they target. Emotions arise in response to their cognitive bases – hence, the prefix re- – but they are responses that have perceptual-like features – hence the suffix -ceptive.

This characterization allows the receptive theory to avoid the main objections raised against the perceptual theory (for details, see Tappolet 2022). Importantly, this move is non-ad hoc. It would be ad hoc if there were no independent reason to think that a state can have non-conceptual content without also being a perceptual state. And indeed, this has been the leading, albeit implicit, assumption in the debate about the nature of emotions for a long time. But it is an unwarranted assumption. As Jacob Beck (2012), for one, has pointed out, pigeons appear to undergo non-perceptual non-conceptual states about magnitudes, such as numbers. Having non-conceptual content and being a perceptual state are thus not co-extensive categories. The receptive theory takes this lesson on board and offers a theory of emotions that preserves the main insights of the perceptual theory, while avoiding its defects. I will assume this theory in what follows.

3.2 The Intentional Object and the Vehicle of Loneliness

If the receptive theory is the chosen approach to emotions, then loneliness must be modeled in accordance with it. Generically, we can say that an episode of loneliness consists in a receptive experience that non-conceptually represents its object as possessing a particular evaluative property. The real task, however, is to give substance to this generic account, so as to meet the desiderata listed above.

I will start by considering the following question: Does loneliness have a specific kind of intentional object? As we have seen, R&K hold that loneliness concerns absent social goods. I think they are basically right in this respect, so I would like to preserve this idea, although with some amendments. To begin with, I think that it is preferable to talk about relational goods, rather than social goods. Some of the goods that are missing in episodes of loneliness, such as intimacy with one’s partner, are not very ‘social’, but they are certainly ‘relational’. The main difference, however, concerns the kind of states that constitute loneliness and that have the relevant goods as their intentional objects. R&K state that loneliness consists in a pro-attitude towards certain social goods and in a non-evaluative awareness that they are out of reach. In my view, instead, loneliness consists in an evaluative receptive state. More specifically, it consists in a receptive state that represents the absence of certain relational goods as being bad in a specific way.Footnote 9 This statement highlights a more subtle difference with respect to R&K’s account. In my view, the intentional object of loneliness is the absence of certain relational goods, rather than the absent goods themselves. It is indeed the absence that is represented as being bad in a certain way, not the relational good that is missing. For example, it is the fact of not having any friends, or no one to talk to, or of not being understood, which is evaluated as bad in an episode of loneliness. This evaluation is at the heart of loneliness, and I will have more to say about it in what follows. For the moment, however, let us keep exploring the differences with R&K’s account.

Insofar as the receptive experience that loneliness consists in has some perceptual-like features, we can say that it involves an evaluative awareness of the absence of certain relational goods. What about the non-evaluative awareness that these relational goods are absent, which was central in R&K’s account? As anticipated in Subsection 2.2.3, I think that this state is, at most, one of the cognitive bases of loneliness. If the term ‘awareness’ is given a broad meaning, so as to include a wide range of cognitive states, then it might even be the case that it is a necessary cause of loneliness. But this is not the same as saying that it is a constituent of it.

What about the pro-attitude towards the absent relational goods? In Subsection 2.2.2, I suggested that the claim that loneliness involves a pro-attitude of this sort is most plausible if the pro-attitude is conceived of as a valuing attitude. After all, we typically feel lonely when we lack friends, because we value having friends. At the same time, I argued that if valuings are conceived as involving either conative states or value judgments, then they turn out not to be necessary for undergoing loneliness. As a matter of fact, however, I think that these are not plausible accounts of valuings. Elsewhere (Rossi and Tappolet manuscript), Christine Tappolet and I have defended the thesis that valuings should be identified with sentiments. It remains to be seen, then, whether valuings thus conceived are necessary for loneliness.

Tappolet and I conceive of sentiments as standing evaluations that dispose an individual to experience tokens of different emotion types. In our view, sentiments have a representational content akin to that of emotions: they non-conceptually represent their objects as possessing specific evaluative properties. Thus, for example, the sentiment of love for a person involves a non-conceptual representation of that person as lovable. This feature is what makes sentiments ‘evaluations’. What makes them ‘standing’ evaluations is the fact that, unlike emotions, they are not occurrent states, but dispositional states. More specifically, they are multi-track dispositions. They dispose an individual to experience a variety of emotions. Thus, for instance, the sentiment of love for their partner may dispose an individual to experience joy at their partner’s promotion, admiration at their partner’s skills, sadness at their partner’s illness, and so on.

According to Tappolet and I, valuings are sentiments thus conceived. Accordingly, valuing intimacy in a relationship amounts to having a positive sentiment (e.g., love, care, like) towards it. The question then becomes: Is it necessary to have sentiments of this sort to undergo episodes of loneliness? For example, is it necessary to have a positive sentiment towards being in a romantic relationship to experience loneliness in the absence of such a relationship? Many instances of loneliness seem to be of this sort. Still, I do not see why it would be impossible to experience loneliness in the absence of the relevant sentiments. Consider an individual who has a lot of friends and is happily single. One day, one of his friends announces to him that he is in a relationship. After seeing his friend with his partner, the individual suddenly realizes that he is lacking something important in his life and feels lonely. I think that this scenario is not implausible. If this is true, then it shows that one need not have formed a standing disposition towards being in a romantic relationship to be able to apprehend their lack of a romantic relationship as bad.Footnote 10 Generalizing, my view is that, for an individual to experience loneliness, it is not necessary that they have a pre-existing sentiment that disposes them to have this experience.

I might be wrong about that. If I am, then valuings as sentiments are necessary pre-conditions for undergoing loneliness and episodes of loneliness are manifestations of the individual’s underlying valuings. It is worth noticing that this account would be immune from the main objection raised against the value judgment account of valuings. Since sentiments do not require the possession of evaluative concepts, this account allows for the possibility that individuals such as children and non-human animals experience loneliness, despite their limited conceptual resources.

I want to conclude this subsection by coming back to the intentionality of loneliness. It seems to me that, in addition to having an intentional object, episodes of loneliness also have what I will call, for lack of a better word, an ‘intentional locus’. Consider the following. An individual who moves to a different city may feel lonely because of their lack of friends even if they have a solid network of meaningful relationships elsewhere. In this case, the individual feels lonely for their lack of friends in the new city. Friendship is the relational good involved in this episode of loneliness. Its absence is the intentional object of the emotion. But this absence is situated in a given context, namely, ‘the new city’. The latter is the intentional locus of the episode of loneliness, where, by ‘intentional’, I mean that it is part of the intentional content of loneliness; it does not just identify the context in which the episode of loneliness takes place. Other examples can easily be given. Individuals may feel lonely within their romantic relationship, within their family, or in the workplace. They miss some goods that are situated in those contexts, and they experience intentionally situated loneliness as a result.

3.3 The Formal Object of Loneliness and the Problem of Demarcation

In this subsection, I want to consider two further questions. What is the evaluative property that instances of loneliness represent their objects as possessing? What distinguishes loneliness from other emotion types? It turns out that these questions are closely related. Before explaining why, let me introduce a bit of terminology.

Following standard use, I will call the evaluative properties that emotions target their ‘formal objects’. On the receptive account, as we have seen, the formal object is part of the representational content of emotions. Other accounts, such as the attitudinal theory (Deonna and Teroni 2012, 2015), deny that. This notwithstanding, most accounts of emotions agree that formal objects play a role in demarcating different emotion types. On the receptive account, this role is spelt out as follows: (at least part of) what identifies an emotion token as belonging to a specific emotion type is the type of evaluative property that it represents its object as possessing. Thus, for example, all tokens of admiration are tokens of the same emotion type because they all represent their objects as being admirable. From this, we can immediately see how the questions above are related: to identify the formal object of loneliness is to identify (at least part of) what distinguishes loneliness from other emotion types.

Before proceeding, let me explicitly emphasize one point. As I understand it here, the problem of demarcating loneliness from other emotions is a metaphysical problem. The goal is to identify the property (or the set of properties) that makes loneliness a distinct emotion type and, thereby, the property (or the set of properties) in virtue of which some emotional tokens count as instances of the emotion of loneliness and not of other emotions. This problem is different from the epistemological problem of how we can recognize, form beliefs, or know that some emotional tokens count as instances of the emotion of loneliness and not of other emotions. In fact, the two problems are independent from each other. It is indeed perfectly possible that the property (or the set of properties) that makes loneliness distinct from other emotions does not coincide with the property (or the set of properties) that we use, or to which we refer as evidence, to recognize or judge that some emotional tokens belong to the loneliness type.Footnote 11

With this in mind, let us go back to the issue of the formal object of loneliness. Loneliness is a negative affective state, a state that appears to involve a negative affective evaluation of the individual’s condition. This suggests that the formal object of loneliness is a negative evaluative property. We can reformulate this by saying, as I have done in the previous subsection, that loneliness represents the absence of certain relational goods as bad in a specific way. Unfortunately, we lack a term to denote the specific way in which the intentional object of loneliness is represented as being bad. In this sense, loneliness differs from most other emotions. Consider two examples. All instances of admiration represent their objects as being admirable. ‘Admirability’ is thus the term that we use to identify the formal object of admiration. Similarly, all instances of fear represent their objects as being fearsome. ‘Fearsomeness’ is the term that refers to the formal object of fear.

Does the lack of a term to designate the formal object of loneliness present a problem when it comes to distinguishing loneliness from other emotions? No. To demarcate loneliness, we must be able to provide a characterization of the evaluative property which loneliness aims at and that its tokens non-conceptually represent. This does not require having a pre-existing term to designate this property. That said, it must surely be acknowledged that characterizing the formal object of loneliness is not an easy task. But in this respect the situation of loneliness is hardly different from that of other emotions. Consider again admiration. To distinguish admiration from other emotions, we need to provide a characterization of the property of admirability. Part of the difficulty in doing so derives from the fact that, conceptually, we have little option but to relate admirability back to the emotion of admiration, by saying, for instance, that the concept of admirability is either identical or equivalent to the concept of fitting admiration.Footnote 12 This does not seem to help much in shedding light on the nature of admiration. After all – we may ask – if we say that admiration is distinct from other emotions because it aims at admirability, but then we need to refer back to admiration to conceptually elucidate admirability, how much progress have we really done on the problem of demarcation?Footnote 13

The crucial thing to notice, however, is that the conceptual account of admirability in terms of fitting admiration is compatible with a metaphysical account according to which admirability is independent from admiration. That is, it is perfectly possible to maintain that although the concept of admirability is a response-dependent concept (in one of the ways envisaged above), the property of being admirable is not. For instance, it is possible to say that admirability is an objective monadic property that certain individuals or objects possess. Indeed, this is the general view about evaluative properties that Tappolet and I have defended elsewhere.Footnote 14 This metaphysical account opens a space between admiration and admirability that allows us to offer a substantive account of the relation between the two. For a start, we can say that the emotion of admiration aims at disclosing the property of being admirable and that admirability is the property that makes instances of admiration fitting, rather than the property identical or reducible to the property of being fitting to admire. We can also further investigate admirability by identifying the properties that make an object admirable (i.e., the grounds of admirability), as well as the things that possess the property of being admirable. By doing all this work, we can meaningfully distinguish admiration from other emotions in terms of admirability, and this despite the close conceptual connection that exists between the two. To reiterate, the key is to keep in mind that the problem of demarcation is a metaphysical problem, not a conceptual problem. This means that if what distinguishes admiration is the fact that it aims at a specific evaluative property, i.e., admirability, then, if this property is independent from admiration and if we can offer a substantive characterization of it, we have successfully addressed the problem of specifying what distinguishes admiration from other emotion types. This remains true even if the concept of admirability is not independent from the concept of admiration.

The situation is essentially the same in the case of loneliness. The only difference is that we lack a term that plays an equivalent role to ‘admirability’, unless we are inclined to neologisms and are prepared to say that loneliness targets the ‘lonely-worthy’. In light of what we have just seen, however, this is not a problem for demarcating loneliness from other emotions insofar as we can investigate the formal object of loneliness in much the same way in which we can investigate the formal object of admiration. As before, we can say that the formal object of loneliness is the evaluative property that loneliness aims to apprehend and that makes instances of loneliness fitting. We can also attempt to identify the properties that ground the formal object of loneliness and the kind of relational goods whose absence bears this negative evaluative property. This is, of course, a task whose completion exceeds the scope of this paper. It is the task for a substantive theory of value.

That said, even the limited understanding of the formal object of loneliness provided here is sufficient to distinguish loneliness from other emotions of absence, such as social anxiety. I claimed before that the latter involves a kind of fear of not being liked or appreciated. On this understanding, the formal object of social anxiety is similar to the formal object of fear, which is typically identified with the fearsome. Accordingly, we can say that social anxiety is fitting if and only if not being liked or appreciated by other people is really fearsome in the kind of way associated with social anxiety. Note, however, that the property of being fearsome is different from the property of being ‘lonely-worthy’. This is sufficient to demarcate social anxiety from loneliness.

In addition to distinguishing different emotion types from each other, it is also possible to distinguish different forms, or sub-types, of the same emotion type. In some cases, we can identify these sub-types by further specifying their formal object. For instance, I argued elsewhere (Rossi manuscript) that what distinguishes different sub-types of the emotion of interest is the particular way of being interesting that their tokens represent. In other cases, however, what demarcates different sub-types is not their formal object. Consider unrequited love, which R&K mention as another example of emotion of absence. Unrequited love is a sub-type of love. What distinguishes unrequited love from other forms of love is not its formal object, but its intentional object. Indeed, love counts as unrequited if it is directed to a (perceived, believed or supposed) unreciprocating individual. In yet other cases, what demarcates different sub-types of the same emotion is their cause. Prinz (2007), for instance, has argued that what distinguishes ordinary anger from indignation is that indignation is specifically elicited by an injustice. These considerations open new possibilities to enrich our conceptual understanding of loneliness.

The suggestion I want to make here is that we can distinguish different sub-types of loneliness in terms of their type of intentional object. Consider the following example. When an individual feels lonely because they have no friends, the intentional object of the individual’s loneliness is their lack of friends. This object can, however, be redescribed at more abstract levels. Depending on the circumstances, lacking friends can be seen as an instance of social exclusion, or as a failure of social integration, or as an instance of social disconnection. The individual themselves may experience their lack of friends as such and feel lonely as a result. This example can be generalized. The goods that are relevant for loneliness typically belong to more general categories. Likewise, the absence of these goods can be characterized in terms of more general categories of relational bads.Footnote 15 To illustrate, let me offer a non-exhaustive list of these relational goods and bads.

Togetherness – Separateness.

Connection – Disconnection.

Belonging – Unbelonging.

Companionship – Isolation.

Affiliation – Disaffiliation.

Association – Disassociation.

Social attunement – Being socially out of tune.

Closeness – Distance.

Acceptance – Rejection.

Inclusion – Exclusion.

Familiarity – Alienness.

These relational goods and bads are often associated with particular feelings. For instance, we often talk about the feeling of social connection and the opposite feeling of social disconnection; about the closeness or the distance we feel towards some individuals; about the feeling of exclusion that we suffer in some circles and the feeling of acceptance we undergo elsewhere; about the feeling of alienation and estrangement from our family as well as the feeling of companionship with our friends. These expressions are a little ambiguous. When we say that we feel excluded, we may refer to our perception that we have been excluded or to our experience of the badness of the exclusion. In the former case, the feeling is not valenced. It is a neutral feeling, which may or may not be accompanied by an emotional reaction. If I am excluded from a group that I despise, for instance, I may feel excluded, but not be bothered about it. When we experience the badness of the exclusion, instead, the feeling of exclusion is valenced: it is an unpleasant feeling. I am interested in the latter kind of feelings associated with relational goods and bads. How can we make sense of them?

Experiencing the badness of exclusion has all the hallmarks of an emotional episode. That is, when we experience a valenced feeling of exclusion, it appears that we are experiencing an emotion. But which emotion is it? Is it a distinct, sui generis emotion? Or is it a sub-type of another emotion? I want to suggest that when we experience a valenced feeling of exclusion, we undergo a particular form of loneliness, a sub-type of this emotion that has exclusion as its more general intentional object. Let me clarify. I am not saying that loneliness is the only type of emotion that one can experience in response to a non-valenced feeling of exclusion. In fact, one may feel happy for having been excluded from a racist group or disappointed for having been excluded from a sport team. What I am saying is that the expression ‘I feel excluded’, when it is used in a valenced sense, refers to the kind of loneliness that one experiences when they perceive (or believe) to have been excluded.

I think that something similar can be said about the feelings associated with all the other relational bads in the list, such as the feeling of social disconnection, the feeling of social alienness, the feeling of not belonging, the feeling of social isolation, and so on. And indeed, these expressions are often used in the scientific literature as synonyms of loneliness (see, e.g., Cacioppo and Patrick 2009). If I am right, then these ‘feelings’ identify different sub-types of loneliness, that is, different ways of experiencing loneliness.Footnote 16

3.4 The Phenomenology and the Degrees of Loneliness

The previous discussion provides a nice bridge to the next question: How can we characterize the phenomenology of loneliness? In my view, the phenomenology of loneliness is primarily a matter of its evaluative content. As we have seen, loneliness targets a specific negative evaluative property, i.e., a specific way of being bad. My view is that the way a token of loneliness feels depends on the particular way in which its intentional object is represented as bad. I say ‘primarily’ because the specific type of intentional object that a token of loneliness has – more specifically, the particular way in which the absence of the relevant relational good is represented – may also contribute to the phenomenology of that loneliness token. But what determines the valenced phenomenological character of a loneliness token, i.e., its feeling bad or unpleasant, is the evaluative component of its content.Footnote 17

We can use this general account to clarify the phenomenology of the different sub-types of loneliness identified above. Take an instance of loneliness due to a perception of social exclusion. This instance involves two integrated representations: a ‘neutral’ representation of social exclusion and an additional ‘evaluative’ representation of social exclusion as bad in a specific way. The overall phenomenological character of this instance of loneliness is the result of the ‘feelings’ generated by these representations. On the one hand, there is the non-valenced feeling of social exclusion. This ‘feeling’ is determined by the ‘neutral’ representation of social exclusion. Initially, this non-valenced feeling is nothing but the phenomenal character of the perceptual experience of social exclusion that provides the cognitive basis for the emotional experience of social exclusion. But when the ‘neutral’ representation of social exclusion becomes part of the ‘receptive’ experience, then the non-valenced feeling of social exclusion that this representation generates constitutes the phenomenal character of the non-affective component of the emotional experience of social exclusion. On the other hand, there is the valenced feeling of social exclusion. This is determined by the ‘evaluative’ representation of social exclusion as bad in the specific way characteristic of loneliness. This valenced feeling is the phenomenal character of the affective component of the emotional experience of social exclusion. It is the way in which social exclusion feels bad as a result of being represented as bad in that particular emotional instance. A similar description applies to other sub-types of loneliness.

These remarks provide the basis for an account of the phenomenal intensity of loneliness, that is, of the degree to which loneliness feels bad or is unpleasant. Recall that, according to R&K, the phenomenal intensity of an episode of loneliness depends on the strength of the pro-attitude towards a specific social good and, perhaps, on how unattainable the individual perceives that good to be. In my view, instead, the degree to which a token of loneliness feels bad, or is unpleasant, depends on the degree to which its intentional object is represented as bad in the specific way characteristic of loneliness. A bit more formally, we can say that a token of loneliness directed at X, where X is the absence of a relevant relational good, feels bad to degree m if and only if it represents X as bad in the specific way characteristic of loneliness to degree m, where it is understood that degrees of badness need not be numerical degrees or especially precise degrees.

What about the pro-attitude and the non-evaluative awareness of a missing relational good? I think that they can still play some role. For a start, if an individual has a strong pro-attitude towards friendship, they are typically more likely to experience loneliness in its absence. Moreover, the strength of the individual’s pro-attitude may certainly influence the phenomenal intensity of the corresponding episodes of loneliness. Likewise, the degree to which the individual deems the absent relational good to be unattainable may have a causal impact on how bad the individual experiences its absence to be. As I suggested above, however, I think that it is possible for an individual to undergo loneliness without having any occurrent or standing attitude of this kind. After all, there is such a thing as a sudden realization that not having a romantic partner sucks. It seems to me that at least some instances of loneliness are of this sort.

The individual’s valuings, in particular, play another important role. There is indeed another gradable dimension in terms of which instances of loneliness, as well as other affective states, can be evaluated, namely, their centrality. The general idea is that, quite independently of its phenomenal intensity, an affective state may be more or less central in the individual’s mind. For example, the pain one experiences after a cramp may be quite intense, but not especially important in the overall mental economy of the individual. Elsewhere (Rossi and Tappolet 2022), Tappolet and I have distinguished two ways in which an affective state can be central. On the one hand, there is what we called ‘output-centrality’: this is roughly the extent to which an affective state disposes an individual to experience other mental states and perform specific actions, that is, the extent to which that affective state is productive.Footnote 18 On the other hand, there is what we called ‘input (or source)-centrality’: this is roughly the extent to which an affective state is based on other productive states. These categories also apply to loneliness. Its tokens may be more or less central in either of these senses. Clearly, the individual’s valuings are amongst the most productive mental states of an individual, in the sense that they shape much of the individual’s mental life and behaviours. They also play a role in determining how input-central various instances of loneliness are. Suppose, for example, that two individuals experience loneliness because of their lack of friends in a way that is equally phenomenally intense. Suppose, however, that only the former individual strongly values having friends. In this case, we can say that their episode of loneliness is more input-central than that of the other individual.

3.5 Chronic Loneliness

I want to conclude by briefly considering how to characterize chronic loneliness. As we have seen, R&K identify this as an important, yet puzzling phenomenon. They are right: it is indeed the central phenomenon with which scholars working on the ethics and politics of loneliness are concerned (see Cormier manuscript).

I think there are two ways of understanding chronic loneliness. According to a minimalist understanding, chronic loneliness is simply a robust and persistent emotional disposition – more specifically, the disposition to experience episodes of loneliness. According to a maximalist understanding, instead, chronic loneliness is a more encompassing phenomenon, which we can describe as a psychological condition, a category that includes, amongst others, states such as depression. As a condition, chronic loneliness essentially involves a robust and persistent disposition to experience occurrent loneliness, but it is not limited to that. It also involves a range of other emotional and non-emotional dispositions. It is indeed a characteristic of the chronically lonely person to experience negative moods and emotions, to suffer a reduction in their executive control function, to have impaired social cognition, and to adopt various aversive behaviours (see Cacioppo and Patrick 2009). It seems to me that chronic loneliness as a condition is the phenomenon that most interests both social scientists and philosophers.

Independently of how it is exactly understood, it is common to say that chronic loneliness can vary in degree. But this expression is ambiguous. It can be interpreted in the following, non-mutually exclusive, ways. First, it may refer to the fact that chronic loneliness can lead to more or less phenomenally intense episodes of loneliness. Second, it may refer to the centrality of chronic loneliness. In particular, it may denote the dispositional strength of chronic loneliness, that is, the strength with which it disposes an individual to undergo occurrent loneliness as well as other emotional and non-emotional states. Dispositional strength may, in turn, be characterized in terms of the likelihood with which these states manifest or in terms of their activation threshold. Third, it may refer to the resilience of chronic loneliness, that is, to the extent to which the latter is capable of withstanding attempts to eliminate or reduce it or the ease with which it returns to its baseline after more favorable experiences.

I think that all these dimensions are important both for descriptive and normative purposes. The first explains why chronic loneliness is so phenomenally awful when it manifests itself. The second helps predict and explain much of the individual’s behaviour and frame of mind. The third helps explain why it is typically so difficult to get rid of chronic loneliness, e.g., why we cannot tell a chronically lonely person to just go out and meet some people and everything will finally be ok. It is important to investigate all these dimensions in future research.

4 Conclusion

Loneliness is a complex phenomenon. In this paper, I argued that we should reject Roberts and Krueger’s ‘desire-belief’ account of loneliness and adopt instead a receptive account. To summarize, the main features of my account are the following. Loneliness as an occurrent emotion consists in an affective ‘receptive’ experience. It has the absence of a relational good as its intentional object. In an episode of loneliness, this absence is non-conceptually represented as being bad in a particular way. An emotional token of loneliness may, but need to, be a manifestation of the individual’s valuing the relevant relational good, where ‘valuing’ something is here conceived as being identical to ‘having a positive sentiment’ towards it. What distinguishes loneliness from other emotions is the particular way of being bad that loneliness aims at apprehending and that its tokens represent. The valenced phenomenal character of a loneliness token is determined by this evaluative representation: a token of loneliness feels bad, or is unpleasant, in a specific way because it represents the absence of a relational good as bad in a particular way. Chronic loneliness is a more complex state that disposes the individual to experience (amongst other things) occurrent instances of the emotion of loneliness.