Feeling Responsible: On Regret for Others’ Harms

Abstract

This paper investigates the moral emotion of being socially, but non-agentially connected to a harm. I propose understanding the emotion of an affiliated onlooker as a species of regret called ‘social-regret’. Breaking from existing guilt- and shame-based accounts, I argue that social-regret can be a fitting, expressive, and revelatory reactive attitude that opens the way for deliberation over accountability for others’ harms. When we feel social-regret, our attention is directed towards the moral salience of our social relations and the expectations that undergird them, as well as possibilities for ameliorative action. I consider several existing accounts of affiliated onlookers’ emotions (including embarrassment, guilt, and shame), and I highlight the advantages of supplementing these with a regret-based account. Social-regret provides a novel way to understand negative, self-directed emotions in response to others’ harms as rational, expressive, and potentially reason-giving experiences.

1. Introduction

In The Theory of Moral Sentiments, Adam Smith raises the following maxim which he takes to be self-evident: generally speaking, one is only deserving of praise or blame on account of actions for which one is morally responsible (Smith, 2002, p. 109 [II.iii.intro.3]). According to this principle, we should not blame others for contingent events, accidental harms, or unforeseen or unintended outcomes. Smith is also keenly aware, however, that our moral sentiments do not always adhere to this rule. He identifies two notable irregularities. The first is that we tend to decrease sentiments of praise or blame for intended, but incomplete actions; that is, we ‘diminish our sense of the merit or demerit of those actions which arose from the most laudable or blamable intentions, when they fail of producing their proposed effects’ (2002, p. 114 [II.iii.2.1]). Second, we also tend to enlarge sentiments of praise or blame ‘beyond what is due to the motives or affections from which they proceed’ in light of outcomes (2002, p. 114 [II.iii.2.1]). In other words, our moral sentiments are susceptible to seemingly irrational escalation and diminution on the basis of resultant or consequential luck – how things turn out.¹

Smith notes that even those who are unaffected by and uninvolved in an action are liable to experience these heightened moral sentiments: ‘Nor is this irregularity of sentiment felt only by those who are immediately affected by the consequences of any action. It is felt, in some measure, even by the impartial spectator’ (2002, p. 114 [II.iii.2.2]). He goes on to explore several powerful, self-directed moral sentiments that respond to significant relations and group memberships (such as feelings of national pride and shame), and observes that the intensity of these self-directed moral sentiments is often disproportionate to one's own causal contributions to the outcomes in question.²

Over two hundred and fifty years later, the question of moral emotions by association rose to prominence after World War II in discussions of German guilt, and continues to occupy centre stage in contemporary debates over identity, legacies of oppression, environmental degradation, and divestment. How am I to feel when my close family member inflicts harm through bigoted behaviour? What about my neighbour, or fellow citizen? How should I feel regarding the actions of my ancestors? Many people hold strong views about the appropriateness (or inappropriateness) of self-directed moral emotions – guilt and shame, especially – for the wrongdoings of others. One can imagine two extremes: on the one side, the dismissal of self-directed, negative emotions for anything other than one's own conduct as pathological; and on the other side, excessive and unproductive feelings of guilt and shame for events far beyond one's own control.

The object of this article is to reframe the negative emotions of affiliated onlookers in order to find a middle way between these two extremes. My basic claim is that some of the moral sentiments associated with harms brought about by one's social relations are not curiously persistent pathologies, but instead fitting reactive attitudes with morally and politically salient epistemic and conative outcomes.³ Taking these affective responses seriously, we might ask: What does the emotion of an affiliated onlooker express? Where does it pull our attention? How does it predispose us to behave?

The moral emotion that I have in mind is experienced by an affiliated onlooker – that is, someone who is a third party to a harm and does not satisfy standard conditions for moral or collective responsibility, but nonetheless feels implicated in that harm in virtue of her social connection with the harm's perpetrator. Existing scholarship tends to analyse the emotions of affiliated onlookers in terms of guilt and shame (May, 1992; Gilbert, 1997; Jaspers, 2001; Oshana, 2006; Christensen, 2013), most recently in Amy Sepinwall's defence of the rationality of ‘faultless guilt for a loved one's wrong’ (Sepinwall, 2022, p. 210). Supplementing these accounts, I aim to elucidate the emotion of social proximity to wrongdoing in terms of regret in the tradition of Bernard Williams’ concept of agent-regret. In doing so, I hope to shed light on a different dimension of the affective phenomenon in question than is captured in guilt- or shame-based accounts, and to draw wider conceptual distance between the emotions of affiliated onlookers and the emotions of perpetrators.

Here is my basic proposal: when one's social relations and/or (members of) one's social groups behave badly, one may be susceptible to a species of regret that I call ‘social-regret’. Social-regret provides a new name for what Marina Oshana calls the ‘phenomenological experience of [moral] taint’ (Oshana, 2006, p. 364). We feel social-regret when we are not agentially involved in harmful actions, but rather connected through significant social relations to another's harm. To lay the groundwork for a regret-based account, I will begin by describing a scenario in which an individual is connected to a harm purely by affiliation, and in which regret of some kind is a plausible response. I then situate social-regret in relation to other forms of regret and identify three typical features of central cases. Next, I argue that social-regret has notable epistemic and conative outcomes – specifically, it can draw attention to our social relations and the expectations that undergird them, and open the way for novel deliberation over accountability going forward. Finally, I conclude by considering several existing accounts of the emotion of affiliated onlookers: embarrassment, guilt (of various kinds), and shame (of various kinds). I give both conceptual and practical reasons for favouring a regret-based account for the cases under consideration, and I end with the suggestion that social-regret better avoids connotations of culpability where guilt- and shame-based accounts tempt them rhetorically, and so may help to head off common defensive dismissals of the emotion in question as a symptom of a pathologically bleeding heart. Because the language of regret more clearly grants premises of non-culpability and non-agency, regret may be a more attractive conceptual resource for enjoining others to be receptive to their moral emotions and to take accountability for harms that are not, strictly speaking, their own.

2. Between Spectator and Agent

It has long been recognized that emotions can redirect our attention by influencing the perceived salience of information about ourselves (de Sousa, 1987; Elgin, 2007; Roberts, 2013) and those who fall within our ‘circle of concerns’ (Nussbaum, 2013, p. 11), or the community and social environment(s) in which we are enmeshed. My suggestion is that social-regret refocuses our attention on the moral salience of our social relations and affectively opens space for deliberation over accountability.

Consider the following situation:

In her high school history class Martha learns about the American deployment of atomic weapons during World War II. She is viscerally disturbed by the sheer scale of devastation wrought by the bombings, as well the fact that it was her own country that carried them out. When Martha brings up the subject with her parents, they tell her that her maternal grandfather (now deceased) was an engineer who worked on the Manhattan Project. His contributions to the wartime effort to develop and weaponize atomic energy were, as far as her parents understand it, indispensable. Martha's affective discomfort heightens. She already felt somewhat connected to and unsettled by the bombings on account of her nationality, but now she feels that she has a different and alarming relationship to them. Recalling her teacher's arguments that the use of atomic weapons may have saved lives on the Pacific front, Martha tries to rationalize that her grandfather may not have done anything blameworthy, and that his actions were his own, not hers. Even so, the uncomfortable feeling persists. Her cognizance of her grandfather's actions presses upon her.

Some would describe this as a case of moral taint, in which one's moral record, personality, and/or psyche are marred by the conduct of associated parties (Oshana, 2006; Appiah, 1991; May, 1991). There is wide disagreement over whether and how moral taint relates to moral responsibility.⁴ For the purposes of this paper, I will set the question of whether and when moral taint generates moral obligations on the part of affiliated persons to the side in order to focus on the way that moral taint can be experienced through reactive attitudes.⁵ This is to ask where the emotion of vicarious harm – e.g., learning of a family member's harmful prejudices, or a workplace's unethical investments, or a friend's cruel decision – pulls us epistemically and conatively.

Martha knows that she is obviously not to blame for the deployment of atomic weapons long before she was born. Nevertheless, she feels connected to the bombings in at least two ways: first (and faintly) as an American citizen, and second (and more urgently) through her grandfather. Given her distinctive positionality vis-à-vis the harms associated with the bombings, what kind of moral emotion is it appropriate for Martha to feel? I do not mean ‘appropriate’ in the normative sense of how Martha ought to feel. I follow Deborah Tollefsen's position that ‘[t]he notion of appropriateness in discussions of the emotions is one of rational acceptability rather than morally obligatory or compulsory’ (2006, p. 225). Put another way, we can understand the appropriateness of an emotion in terms of what D'Arms and Jacobson call the ‘fittingness’ of an emotional response, which refers to the harmony between evaluative features of an emotion and its object (D'Arms and Jacobson, 2000). In what follows, I take it as given that there are criteria under which certain moral emotions are fitting in this evaluative sense, and that the absence of those conditions can render affective responses unfitting.⁶

A starting point can be found in the category of ‘counterfactual emotions’, a term that Kahneman and Tversky use to denote a class of emotions that are dependent ‘on a comparison of reality with what might or should have been’ (1982, p. 206). Regret is a counterfactual emotion in that it involves reference to alternative possibilities. Note that one need not have a specific alternative reality in mind when experiencing regret. Counterfactuals might simply negate certain elements of a state of affairs; for example, one can regret going to college without having a clear sense of what one would have preferred to do instead.

It is plausible that Martha might appropriately feel regret for the actions of the United States military despite her lack of agential contribution. In its most general sense – what Williams calls ‘regret in general’ (Williams, 1981, p. 27), or what Carla Bagnoli calls ‘evaluator-regret’ (Bagnoli, 2000, p. 176) – one can feel regret for practically any harm, including those that took place before one was born.⁷ Still, although it is true that evaluator-regret is fitting for Martha's situation since it ‘can in principle apply to anything of which one can form some conception of how it might have been otherwise, together with consciousness of how things would then have been better’ (Williams, 1981, p. 27), it seems far too general to say that Martha feels the regret of a mere onlooker or spectator. As R. Jay Wallace notes, ‘[u]ninvolved third parties do not have the same special reasons for regretting… [and] they can only feel a more generic pain or distress that it is in principle open to anybody to experience’ (2013, p. 39). Martha is not merely an onlooker, like a random passer-by who happens to witness a car accident. Her relationship with her grandfather means that she is something more than a spectator, albeit something other than a perpetrator.

Williams famously argues that we can rationally feel ‘agent-regret’ for harms that we bring about non-culpably. To take Williams’ example, a lorry driver who injures a child in an unlucky and non-culpable accident may fittingly experience a unique kind of regret as a result of his agential relation to that accident – that is, ‘there is something special about his relation to the happening, something which cannot merely be eliminated by the consideration that it was not his fault’ (1981, p. 28). In Bagnoli's words, ‘the distinction between agent-regret and evaluator-regret is primarily drawn in terms of subject matter: the agent regrets his actions or his deliberation, while the evaluator regrets some state of affairs brought about by somebody else’ (2000, p. 176). But unlike Williams’ lorry driver, Martha is not agentially connected to the harm in question ex hypothesi. Agent-regret is a consequence of first-person, agential involvement in a harm for which one is not culpable, and for this reason it is unfitting for Martha's situation.⁸

Social-regret falls somewhere between these first- and third-person perspectives. I propose that we think of Martha as an ‘affiliated onlooker’ so as to underscore that her regret has a different character than that of an agent or unaffiliated spectator both. Here are three typical features of the kinds of cases under consideration: (1) one knows and is aware that a harm has occurred; (2) one knows and is aware that the harm in question is reasonably attributable to the actions of a social affiliate (individual, social group, institution) and not reasonably attributable to oneself; and (3) one knows and is aware of the affiliation between oneself and that social affiliate. To apply this to Martha's case, we can say: Martha knows and is aware that (1) the dropping of the atomic bombs brought about great harm, that (2) the harm is in some way attributable to her grandfather's contributions, and (3) that she is affiliated with her grandfather.

These are not necessary or sufficient conditions. I am sceptical of the idea that one can be morally obliged to feel specific emotions, though I concede that there are circumstances in which it would be strange (even abhorrent) for an individual to not feel any sort of counterfactual emotion whatsoever. These features are simply meant to clarify the conditions under which the distinctiveness of social-regret (as compared to other forms of regret) is most clearly discernible. (1) holds that a harm has occurred. Not all regret involves harm; regret in its most general sense is appropriate in almost any case in which one would prefer to live in a world in which things turned out differently in some respect. For example, take the case of an envious restaurant diner who regrets her order after catching sight of a neighbour's meal that looks much more delicious than her own. It would be an exaggeration to say that a harm has occurred. Rather, the diner simply wishes that she lived in the world in which she had ordered the fish. I will set such benign situations aside.

(2) specifies that the harm is reasonably attributable to an individual or social group, broadly conceived. On this point, it might be objected that social practices, as opposed to social affiliates, are the proper objects of social-regret. This is Howard McGary's approach, for example, in his discussion of the conditions for vicarious liability (1986, p. 158). But shared social practices imply the existence of a social group that shares those practices, so long as we include implicit, informal, and unstructured groups in our understanding of what a social group is.⁹ In other words, a capacious understanding of social groups will not overlook harms generated by shared practices, and so (2) refers to social affiliation broadly construed.¹⁰

(2) also specifies that the harm is not directly or reasonably attributable to the actions of the subject who experiences social-regret. This means that the harm in question is not a reasonably predictable outcome of an individual's behaviours, interests, or desires. The set of core cases of social-regret is narrower than the set of cases involving non-agential or remote causation. A cocaine buyer may be a non-agential, distal cause of violence related to systems of drug production and trafficking, but that violence is a predictable outcome of the buyer's desires, interests, and behaviour, whereas social-regret is fitting when there is no such link. Consider, in contrast to the cocaine buyer, an athlete on a sports team who discovers that her teammates are doping. Though it could be said that the athlete's teammates took performance enhancing drugs for the sake of her interests, or that the athlete's desire to win played a causal role in her teammates’ decisions to dope, the harm in question is not a straightforward expression of her interests, desires, or behaviours, since cheating is not fundamental (and is perhaps anathema) to group membership on a sports team.

Finally, (3) accounts for the subject's feeling of being connected to or associated with the harm. When one feels social-regret, one typically considers oneself (at least implicitly) to be affiliated with the relevant social group or individual. If, for some reason, one is unaware of such an affiliation, social-regret is unlikely to occur. One complication is that sometimes ignorance of social affiliation is actively maintained, as Charles Mills argues in his analysis of white ignorance (Mills, 2007). For example, in Ashley W. Doane's words, ‘whites tend not to see themselves in racial terms and not to recognize the existence of the advantages that whites enjoy in American society’ (Doane, 2003, pp. 13–14). If one refuses to acknowledge specific facets of one's social situatedness or identity, then one may avert social-regret even after learning of the harms caused by an affiliated party. (3) suggests that the experience of social-regret can be evaded by denying or maintaining ignorance of one's affiliations with a given harm.¹¹

These three features can also be used to explain the relationship between evaluator-regret, agent-regret, and social-regret. An individual feeling evaluator-regret typically satisfies (1); an individual feeling agent-regret typically satisfies (1) and a modified version of (2), such that the harm in question is reasonably attributable to the subject herself; and an individual feeling social-regret typically satisfies all three conditions.

3. What Does Social-Regret Do?

So far, I have described some of the basic features of social-regret and situated it among other forms of regret. But what does social-regret do? That is, what potential insights and practical opportunities arise from experiences of social-regret?

Social-regret, like other moral emotions such a guilt, shame, and indignation, can be an expressive and revelatory experience. The counterfactual reasoning inherent to regret means that it can be understood as ‘a way of expressing concern and of conferring value’ (Bagnoli, 2000, p. 169). More specifically, social-regret expresses that one values unrealized alternatives, and it may shed light on the principles implied by those alternatives.¹² It can also alert us to the existence of implicit, but previously unarticulated in-group expectations that have been violated. Wallace argues that a key characteristic of reactive emotions that sets them apart from other attitudes is their close connection to the expectations to which we hold others. ‘[T]o hold someone to an expectation’, Wallace writes, ‘… is to be susceptible to a certain range of emotions if the expectation is violated, or to believe that it would be appropriate for one to feel those emotions if the expectation is violated’ (Wallace, 1994, p. 23). In a similar way, we are susceptible to social-regret because we hold and are held by others to normative expectations.

To be sure, we are not always conscious of the details of the expectations to which we hold others or to which we are held, and we may even hold inaccurate views about them. Emotions can intervene on such ambiguity by ‘rendering previously ignored features and previously unknown patterns salient’ (Elgin, 2007, p. 45), and ‘provid[ing] orientations that render particular facets of things salient’ (2007, p. 33). Someone who discovers new and unexpected feelings of romantic jealousy when a friend begins dating someone else might be prompted to realize that the story she tells herself about their relationship is untrue. Here, her feeling belies her self-perception. Just as such a person might ask herself, ‘Do I feel jealousy because I am in love with my friend?’, an individual experiencing social-regret is well positioned to ask after the details of a harm and the moral salience of the social relation in question – for example, whether the harm evinces a violation of an in-group expectation or self-applied commitment.

A natural question is whether social-regret alerts us to real reparative obligations. At this stage, however, it is important not to expect too much of the emotion. Social-regret is not sufficient for (or even a reliable indicator of) ameliorative duties. But it does influence our attention and behaviour in ways that open space for novel deliberations over accountability. For the purposes of introducing the concept of social-regret, I will remain neutral on the question of the content and grounding of the potential ameliorative duties and accountability practices that might come into view via social-regret in different contexts.¹³ We can speak to the epistemic and conative outcomes of social-regret – that it affectively prompts us to learn more about the situation that gave rise to it, and to reflect upon one's relationship to the person or group on account of whom one experiences it – even without committing to a specific account of concomitant responsibilities.

Ascribing these outcomes to social-regret does not mean that everyone who experiences it will actually weigh the moral salience of their social connection to a harm. Like anger, pride, guilt, jealousy, and other emotions, social-regret can be suppressed, ignored, denied, deferred, or drowned out by other affects, emotions, desires, projects, commitments, and obligations. As Myisha Cherry argues, the action tendencies of emotions – that is, the ‘behavior[s] that a person is likely to engage in, given the [emotion]’ (2021, p. 11) – are defeasible, but nonetheless carry moral and political stakes:

Emotions motivate us to act in a certain way. They can also influence our beliefs and desires. Yes, we can act or not act, and the emotion can only do its thing through us and in partnership with us. If I say the action tendency of anger is to approach a target, I simply mean that the anger motivates us to do so. What we eventually do is up to us. (Cherry, 2021, p. 14)

This is a helpful way to think about social-regret. The action tendencies of social-regret are those of investigation and reflection upon social relations and relata, and these arise out of affects of pressure, discomfort, attentiveness, alarm, and persistent curiosity, like that of rubbernecking drivers who cannot help but look at an accident. It is through this shift in attention that significant moral and political outcomes (including, perhaps, the taking on of responsibility and/or accountability for others’ harm) are made possible.

4. Why Regret?

In what remains I will argue that there are both conceptual and rhetorical reasons for analysing the emotion of social proximity to harm in terms of regret over (or alongside with) existing accounts of embarrassment, guilt, or shame.

4.1 Embarrassment

It is first helpful to distinguish between social-regret and embarrassment. It is true that in some circumstances the actions of social affiliates cause embarrassment. Consider, for example, Gabriele Taylor's account of embarrassment at the behaviour of a fellow countryman:

[A] countryman of mine behaves badly (in my view) in front of a foreign audience. I feel embarrassed. Why? He is being seen as a poor specimen, or so I imagine. Because of our shared nationality this verdict affects me as well. But I object to being so seen, so in some way I have to dissociate myself from my countryman, or correct the impression. Here the tension arises not because I actually identify with him, but because I believe that such identification is forced upon me by the audience on the basis of our connection. (Taylor, 1985, p. 72)

Several of the facts in Taylor's case fit the profile of social-regret: she sees that (1) a harm has occurred, that (2) the harm is done by a social affiliate and not herself, and that (3) she is affiliated with the actor in some way. Note, however, the importance of the audience: Taylor's embarrassment arises from the fact that her countryman is seen by an audience who also sees (or might see) that the two are affiliated. ‘In embarrassment’, Taylor goes on to say, ‘concern is always with one's own position vis-à-vis others’ (1985, p. 75). In this case, identification with the foolish actor is ‘forced upon’ her by the thought that the audience could (hypothetically) turn their gaze in her direction and see her in an undesirable light.

This is not the concern that underlies social-regret, which can occur without reference to the spectre of a seeing audience, actual or imaginary. This is not to deny that embarrassment is sometimes a fitting response to the wrongdoings of social affiliates, but only to assert that there is a separate affect of social interconnection with wrongdoing that is not reducible to concern for being identified with a wrongdoer by others. The basic case for social-regret is this claim: that there is an affective weight of a social affiliate's wrongdoing apart from concerns about being seen to be affiliated. Martha, for example, need not expect that others will see her in a negative light because of her grandfather when she experiences social-regret for his actions. Of course, there may be specific situations in which they would, or in which Martha becomes especially attuned to others’ attention. Were Martha to travel to the Hiroshima Peace Memorial, for example, she might understandably be apprehensive about sharing her family's history with descendants of those directly affected by the bombings, or feel broadly conspicuous for being affiliated with the nation that enacted them. But such interpersonal awareness is not the core of her initial emotional reaction to learning of her grandfather's contributions to the development of nuclear weapons, which consisted simply of a felt sense of social proximity to harm. Put another way, social-regret is not a kind of dismay over the implications of another's harm for one's own standing.

4.2 Guilt and Collective Guilt

Insofar as guilt involves holding oneself to a demand and judging that one has fallen short of that demand (Wallace, 1994), it follows that guilt (understood as a self-directed negative assessment of one's own conduct) is not an appropriate moral emotion for cases of non-agential social proximity to wrongdoing.¹⁴ Still, one might turn to guilt-like emotions by way of what Tollefsen calls ‘collective emotions’, which are felt ‘in response to the actions of groups’ (2006, p. 222). Collective guilt has been argued to fittingly arise out of a variety of connections to wrongdoing, ranging from situations in which a single in-group member causes harm to those in which the collective qua collective is morally responsible. Some have even proposed that it is fitting to experience guilt in situations in which an agent merely benefits from ongoing group-based inequality (Branscombe and Doosje, 2004, p. 6), and Christensen coins the term ‘innocent guilt’ to refer to the keen awareness of the ethical claim generated by the suffering of others to which one has not contributed (Christensen, 2013). Several of these more capacious concepts of collective guilt appear to be fitting moral responses for someone in Martha's situation.

These forms of guilt are broader than the subset of cases at issue here. Collective guilt is not only appropriate in cases of moral taint by mere association, but can also fittingly arise out of a diversity of agential linkages between a group member and a harm, such as when an individual has failed to prevent a harm (Lickel, Schmader, and Barquissau, 2004, p. 42), benefits from a harm (Iyer, Leach, and Pedersen, 2004, p. 266), or is a member of a collective that is culpable qua collective (Christensen, 2013, p. 369). In short, collective guilt is not restricted to purely social affiliation to wrongdoing, while my goal is to demarcate the distinctiveness of the emotion that is fitting when an individual is connected to a harm solely in virtue of social relations, and where that individual does not otherwise bear individual or collective liability.

Margaret Gilbert's analysis of ‘membership guilt’ comes close to capturing the specific moral emotions of affiliated onlookers in terms of guilt. Gilbert and Priest argue that it can be appropriate for members of a plural subject (which Gilbert elsewhere uses interchangeably with the term ‘social group’¹⁵) to feel membership guilt in response to the misconduct of their group, even when personal guilt is inappropriate (Gilbert and Priest, 2020, p. 32). Since plural subjects are sets of jointly committed persons, membership guilt can be considered to be ‘self’-directed towards the plural subject, as opposed to the individual self. To similar effect, Sepinwall argues that ‘faultless guilt’ can fittingly ‘emerge from a sense of shared agency that partly constitutes their relationship with the wrongdoer’, and that we ought to affirm and ‘perhaps even support efforts to enact the moral and material responses that [experiences of faultless guilt] produce (e.g., offers of repair)’ (Sepinwall, 2022, p. 215). By grounding her account of faultless guilt in shared agency, Sepinwall is able to draw connections to moral emotions precipitated by diffused agential connections to complex systems that produce harm, such as structural injustice and white supremacy.

Although Gilbert and Sepinwall's respective analyses are highly convincing, an important reason for seeking out an alternative account that takes regret, not guilt, as its genus is that Gilbert's concept of membership guilt and Sepinwall's concept of faultless guilt do not apply to social affiliates that are not members of plural subjects with shared agency.¹⁶ A regret-based account is able to accommodate cases that involve social affiliates with whom joint action or shared agency is impossible. This is readily discernible in cases of social-regret for the conduct of now-deceased ancestors. Martha, for example, experiences social-regret with regard to the actions of her grandfather, with whom she does not (and could not, for the time period in question) share agency. Understanding this case through social-regret allows us to see Martha's moral emotion as akin to, but distinct from the moral emotions of harms involving joint agency.

4.3 Shame

According to Helen Block Lewis, the distinction between guilt and shame lies in the fact that the object of guilt is one's behaviour, while the object of shame is one's self, and this distinction has been enormously influential in philosophical conversations around moral emotion (H.B. Lewis, 1971).¹⁷ Shame is a plausible moral emotion for Martha because, like regret, it can be appropriate when an individual is non-culpable, lacks an agential connection to the harm in question, and is connected to the harm through group membership. For example, May argues that shame is an appropriate affective response to moral taint associated with group-level wrongs or faults that result in collective omissions (May, 1992, p. 121), and Oshana argues that ‘the [morally] tainted person would feel shame if she felt as she ought to feel’ (2006, p. 364).

Still, I worry that shame does not adequately capture situations like Martha's. Let us distinguish two facets of shame: its ‘value-oriented’ dimension and its ‘other-oriented’ dimension (Laing, 2022). The former has to do broadly with taking oneself to be deficient with regard to a norm, standard, or expectation, while the latter has to do broadly with one's concern with others’ ‘evaluatively laden psychological states of which one is the object’ (2022, p. 235).¹⁸ Depending on the examples and accounts at hand, sometimes shame involves a feeling of ‘fall[ing] short of a standard of excellence’, and sometimes it involves a feeling of being viewed in a negative light, and/or apprehension about being ‘cut off and rejected, made an object of scorn and ridicule’ by others, and sometimes a mixture of both (Rawls, 1999, p. 423).¹⁹

Cases of social-regret do not fit either picture. To take a value-oriented approach, the moral emotions of social connection to wrongdoing need not involve feeling or thinking ‘that one falls short of some value or standard’ (Laing, 2022, p. 236); nor, by an other-oriented approach, need they involve losing face in the eyes of others or the threat of suffering social avoidance. Martha, for example, does not feel that she falls short of a standard, and (as discussed above in the discussion of embarrassment) does not anticipate being seen in a negative light by others. Her emotion is neither self-evaluative nor oriented towards others’ psychological states. Even Laing's recently proposed middle way between the value-oriented and other-oriented poles of shame, according to which shame involves ‘feel[ing] oneself to merit social avoidance or rejection by others’, does not fit the emotion of purely social proximity to wrongdoing (2022, p. 250). Martha does not take herself to merit social avoidance; rather, she feels herself to be inordinately connected to a harm partially brought about by someone close to her.²⁰

One might nevertheless suspect that social-regret falls under the umbrella of shame, understood in its most general sense as simply the feeling of being distressingly connected to something undesirable. I agree that under this expansive definition Martha can be understood to feel shame. But consider these additional reasons to favour the regret-based account offered above. The first reason is that under such a broad conception of shame, regret (in general) is also a species of shame, as well as guilt and many other negative, counterfactual emotions. If we say that it is fitting for Martha to feel shame without qualification, then we appeal to a rather broad swathe of our emotional lives and cover over the subtle affective relationship under consideration here.

To avoid this outcome, one might specify an anomalous species of shame – perhaps ‘vicarious shame’ or ‘shame by association’ – just as I have appealed to species of regret (agent-regret, social-regret). Working within the genus of shame, it could be argued, tracks closer to ordinary language. But ordinary language cuts both ways, and this brings us to a second reason to seek an alternative account: shame is often associated with self-disapprobation and/or feelings of guilt in our everyday lives.²¹ Where my focus is on non-culpable and non-contributory relations to another's harm, the language of shame tempts connotations of moral disapprobation – in Taylor's words, ‘[a] person feeling shame judges herself adversely’ – whether related to self-evaluation or consciousness of others’ evaluations of the self, or both (1985, p. 68).²² The emotion I have called social-regret, however, is distinctive because it occurs even when no such self-disapprobation or perceived disapprobation by others is present. What is at stake is not one's own moral or social status, but instead the felt force of an affiliate's deeds.

A related reason to separate the emotions of an affiliated onlooker from shame is that shame is often said to carry troubling affective costs. Many scholars already worry over the tendency for shame to become pathologized and give rise to self-destructive behaviours.²³ If we say that it is fitting to feel shame for the wrongdoings of affiliated persons and groups, then we risk radically multiplying opportunities for fitting shame, thereby exacerbating concerns about overburdening. Moreover, the action tendencies of shame typically involve retreat from social contact.²⁴ Reframing the emotional experience of an affiliated onlooker away from shame is an opportunity to reassess the way in which that emotion makes a claim upon us. My suggestion is that regret, while hardly immune from pathologies, gives us a less morally-charged and affectively demanding way of accounting for the emotions of affiliation with wrongdoing, and that it consists of social action tendencies – as I say above, investigation and reflection upon social relations and relata.²⁵ I will grant that there is an affective resemblance between social-regret and shame, and I do not deny that there are shame-based accounts that can help make sense of the emotions of affiliated onlookers. By thinking through this emotion in other terms, however, we might better prepare ourselves to be moved towards productive social responses to an affiliate's harm.

This brings us to a final consideration in favour of distinguishing social-regret from shame: a regret-based account may, in certain contexts, avoid the tendency for guilt- or shame-based accounts to be dismissed as irrational, pathological, or excessive. Rhetorics of shame and guilt can elicit sharp and unproductive resistance, and it is easy to conflate species of guilt or shame that do not imply fault with species of guilt or shame that do (Young, 2011). Individuals in situations like Martha's may understandably be reluctant to describe their feelings in terms that generally connote fault, liability, or stigma. As Jacquet emphasizes, ‘[s]hame is a painful emotion’ that invites a range of possible defensive reactions (2015, p. 151). Gilbert makes much the same point about guilt: ‘feeling guilt is unpleasant, and may have an inherently punishing character’ (1997, p. 83). Though Gilbert goes on to claim that the unpleasantness of guilt can serve as a ‘stimulus to improvement in group action and the moral quality of group life’ (1997, p. 83), that same unpleasantness can just as easily (and perhaps more frequently) elicit defensiveness, denials, and refusals. Tollefsen illustrates this in her description of Mississippi Senator Thad Cochran's refusal to sign his name to a resolution of apology for the Senate's past failure to pass anti-lynching legislation:

When asked by a reporter why his name was absent from the list of sponsors Cochran said: ‘I'm not in the business of apologizing for what someone else did or didn't do. I deplore and regret that lynching occurred and that those committing them weren't punished, but I'm not culpable’. (Tollefsen, 2006, p. 228)

Note that the senator's resistance to apologize for the harms of the (past) Senate is grounded in his intuition that apologies imply wrongdoing,²⁶ as if to say, ‘Apologies are for wrongdoings, and I have done no wrong, and so I will give no apology’. The associations between apologizing and culpability tempt the senator to conflate being affiliated with a harm with being at fault in a harm, and then to dismiss the moral emotions and associated obligations following from the former as excessive and unfitting in light of the evidence that he does not satisfy conditions for the latter. This defensive reaction against the language of guilt is part of what Karl Jaspers acknowledges when he writes that ‘in a way which is rationally not conceivable, which is even rationally refutable, I feel co-responsible for what Germans do and have done’ (Jaspers, 2001, p. 74). Admittedly, Jaspers is unique in that he does not dismiss feelings of guilt beyond those related to one's individual conduct as irrational, but in this passage he openly admits to the intuition that his own emotions, when described and understood as a form of guilt, are irrational. So long as we are primed to associate guilt with one's own culpable actions and shame with the moral or social status of the self, adjacent guilt- and shame-like feelings for others’ conduct will intuitively appear excessive and irrational.

The senator's reaction illustrates the way in which some might recoil at the idea that they ought to feel guilt or shame for others’ harms, no matter how insistent moral philosophers may be that certain species of guilt and shame do not imply as much. Anecdotally, my own experience is that many non-philosophers and philosophers alike are reflexively dismissive of moral emotions bearing the name ‘guilt’ or ‘shame’ in the absence of personal wrongdoing or character deficiencies. Relying solely on these terms makes it far too easy for some (sometimes in bad faith, sometimes not) to dismiss or suppress feelings of moral implicatedness via social affiliations.

But here is the rub: note that the senator is more comfortable using the language of regret – ‘I deplore and regret that lynching occurred and that those committing them weren't punished’ – to respond to the wrongdoing of others. If our feelings of social proximity to wrongdoing have anything to teach us, as I suspect they do, then it will be important to speak about them in ways that allow us to sit with them long enough to move through them. Drawing a more obvious terminological distinction between, on the one hand, the emotions proper to individuals who are linked to harm solely in virtue of their affiliations, and, on the other hand, those proper to individuals with other morally charged connections to wrongdoing (agency, omission, benefit, shared responsibility, solidarity, and more) can better protect the former from being wrongly judged to be pathological cases of the latter. When we work through negative, retrospective reactions to others’ harm in terms of social-regret, a wider sense of social interconnection with persons, groups, institutions, and legacies comes into view. Harms that would otherwise appear to be another's to repent for and repair may, in some cases, emerge as opportunities for moral deliberation and shared accountability.

5. Conclusion

I have identified social-regret as an affective phenomenon that is distinct from both agent-regret and spectator-regret, and argued that it redirects our attention counterfactually towards valued alternative possibilities, the moral salience of the relation(s) in question, and the details of their harm. We would do well to explore new, alternative ways of describing the emotions of social proximity to harm, and the concept of social-regret is one such alternative. To be clear, this does not mean abandoning previous analyses of moral taint, guilt, or shame, many of which have explanatory value beyond the scope of the cases I focus on here. Instead, my goal is to offer a novel account through the lens of regret that can find its place alongside these existing accounts, that accommodates a larger set of cases than those involving joint agency, and that might circumvent some of the more entrenched moral associations of the language of guilt and shame.²⁷