The phenomenon of structural injustice is beginning to receive a great deal of attention, both inside academia and in society at large. Many philosophers have become convinced that we need to develop a conception of moral responsibility which renders individuals responsible, in some difficult-to-pin-down sense, for harms that accrue as a result of structural forces. Arguably, these harms include those produced by persistent racial inequality, climate change, factory farming and sweatshop labor. It is therefore no wonder that philosophers have begun to search for a theory of responsibility which can provide some guidance on these urgent issues (Aragon and Jaggar 2018). Iris Young’s social connection model is just such a theory, and it has had enormous influence on emerging discussions.Footnote 1

The essential core of Young’s forward-looking model of political responsibility has been embraced or defended by many academics working on structural injustice, both in philosophy and elsewhere (Darby and Branscombe 2014; McKeown 2018; Applebaum 2007; Larrère 2018; Zheng 2018). In this paper, I argue that the model has systematically neglected the key role played by the emotions in human moral life. The result is that Young’s model leaves it mysterious how an individual is to be motivated to act in accordance with their political responsibility. To put the point provocatively, Young’s model is a form of ideal theory; a model of justice which assumes that human beings are very different than they are, and which therefore cannot be used to tackle real-world injustice.

What Is Structural Injustice?

The problem Young sought to address in her work is fiendishly difficult. ‘We lack’, she wrote, ‘good conceptual tools for thinking about individual responsibility in relation to structural social processes’ (Young 2003, p. 26; Scheffler 2002, p. 39). This is worrisome, because a great deal of harm, inequality and dysfunction in our social world is best explained by large-scale structural processes, and not by any particular agent’s intentions (Haslanger 2016).

To illustrate, Young provides her widely discussed example of Sandy, a single mother who is forced to leave her apartment and who must try to find affordable housing in a major metropolitan area (Young 2003). Because down-payments are so high, and because she has used all her resources on items necessary to maintaining her job and raising her children, Sandy faces the catastrophic prospect of homelessness.

If we reflect on Sandy’s general situation, we see that she is embedded in a huge network of social forces which are conspiring against her. Every agent in the city is trying to find the best housing they can afford. Landlords are charging what they think they can get away with so that they can pay their own bills. Civic authorities are trying to balance the need for affordable housing with the need for economic growth and prosperity. In other words, millions of agents are unwittingly working together to create the very system of rewards, incentives, inequalities and constraints which eventually renders Sandy and her children homeless.

Here we meet with the problem: if no individual agent intends this outcome, then who is responsible for it? One tempting response is to invoke collective responsibility, and to say that agents who contribute to structural harms constitute a collective which can be assigned a sui generis form of responsibility (Miloski 2016). Of course, such an approach faces many familiar hurdles. For example, it is very hard to find any hallmarks of moral agency in a ‘group’ which is composed of millions of diffuse, largely unconnected agents with no decision-making procedure or internal organization (Feinberg 1968). Perhaps these hurdles can be overcome, but since Young chooses to focus on the question of how an individual is supposed to think of her own responsibility for structural harms, it is this question which will concern me here.

It will be helpful to consider an example of such an individual. An average comparatively well-off homeowner in Sandy’s city has bought property at market value and is simply trying to maintain or increase its overall value. Their activity contributes to Sandy’s predicament, but they do not want people like her to be homeless. They may be compassionate, kind individuals; but as it stands they are just operating normally under a system of constraints and incentives that they did not choose. That system of constraints and incentives (i.e. the structure) seems to be the culprit, here, and not any individual operating within the system. But how is it possible to assign responsibility to a social structure?

It is at this juncture that theorists generally follow Young in drawing a distinction between two kinds of responsibility, forward-looking and backward-looking (Goodin 1995; Rovane 2014; van de Poel 2011). Normally, when we assign responsibility for a harm to an agent, we may do so in a way that essentially involves looking back at the harm they have caused. Young labels the backward-looking mode of responsibility-assignment the liability model. To say that someone is responsible in this sense is to say

  1. 1.

    that their action was causally connected to an outcome in some significant way,

  2. 2.

    that they acted voluntarily,

  3. 3.

    that the outcome was intended or desired, and

  4. 4.

    that they acted under adequate knowledge of the overall situation. (Young 2003, p. 97)

When the outcome is a harm, we thus identify a wrongdoer, and in doing so we activate a host of reactive attitudes, both in ourselves and in the wrongdoer. We feel and express resentment or anger, we perform speech acts of blaming or admonishing, and we activate a host of social mechanisms designed to punish or exclude the wrongdoer. For his part, the wrongdoer is expected to express shame, guilt or remorse, and is expected to acknowledge or recognize his role in the harm.

Elaborating on Young’s idea, Robin Zheng claims that this backward-looking model

derives from a fundamentally metaphysical and action-theoretic problem concerning what actions count as genuine exercises of agency, because only those can provide legitimate grounds for blaming or punishing a person… the responses that make up such moral appraisal, like the paradigmatic (non-consequentialist) praise and blame and the reactive attitudes… reflect assessments of a person’s quality of character as a moral agent. (Zheng 2016, p. 65).

Young argues that this model is both ‘counterproductive’ and ‘inappropriate’ when we are assessing responsibility for structural injustice. Suppose our homeowners rent their basement suite for market value, thus making it unaffordable for Sandy. Their actions are not significantly causally connected to her homelessness, they neither desire nor intend her homelessness, and they, like virtually all agents in the system, have no knowledge of Sandy or of her situation. They are, as Young says, ‘minding their own business and acting within accepted norms and rules’ (Young 2003, p. 106).

This, then, is the problem of individual responsibility for structural injustice. Our central model of moral responsibility seems to simply let virtually everyone off the hook for structural harm. It appears to many of us as though this cannot be right. Is there a way out?

The Social Connection Model

According to Young, there is. Without abandoning the liability model for more ordinary harms, we may assign responsibility for structural harms in a ‘forward-looking’ sense. When we do this, we focus purely on the possibility of an agent’s helping to prevent future harms. To this end, she proposes a social-connection model of responsibility, which is not meant to replace the liability model, but which, rather, is meant to apply in situations when the liability model simply fails to get any purchase. This model has several basic features.

First, it is essentially forward-looking. When this model finds us responsible, this ‘does not imply finding at fault or liable for a past wrong; rather, it refers to agents’ carrying out activities in a morally appropriate way and seeing to it that certain outcomes obtain’ (Young 2003, p. 119). Second, it instructs us to avoid the responses and attitudes characteristic of the liability model. When we are responsible in the social-connection sense, we are not blameworthy, nor are we expected to feel shame or guilt, nor are we meant to suffer any social consequences as a result of our being held responsible. Third, Young suggests that it generates an obligation to try to act with others in remedying structural injustices. These obligations vary, depending on one’s social position. Those who benefit from structural injustice have particularly strong reason to combat it, as do those with a great deal of social and political power (Young 2003, p. 124).

Here is, I hope, a helpful way to see what is going on, here. Within the liability model, the concept of moral responsibility is not temporally indexical. That is to say, if I judge that I am responsible for taking some future action, I must also judge that I will be responsible when the time comes, and that I will have been responsible for performance or non-performance after the time passes.

But Young’s supplementary concept of responsibility is temporally indexical: its truth-conditions change depending on the time at which the responsibility-judgment is made. Put simply, like the concept ‘tomorrow’, our Young’s concept of ‘responsibility for structural injustice’, never refers to the past. The concept simply ceases to refer when the time for action has come and gone. ‘One has the responsibility always now’ she writes, ‘in relation to current events and in relation to their future consequences’ (Young 2003, p. 92).

Zheng has recently tried to argue against this interpretation of Young, citing Young’s claim that ‘both the liability and social connection models refer both to the past and the future’ (Zheng 2019; Young 2003, p. 108). Unfortunately, she neglects to cite Young’s explicit clarification of this point. The social connection model, Young writes,

does need to look backward in one respect. Understanding how structural processes produce and reproduce injustice requires having an account of how they have come about and operated in the past coming up to the present. Having such a backward-looking account also helps those of us who participate in those processes understand our role in them. The purpose of such backward-looking accounts, however, is not to praise or to blame, but to help all of us see relationships between particular actions, practices, and policies, on the one hand, and structural outcomes, on the other. (Young 2003, p. 109)

Young is clear about the very limited sense in which her model is backward-looking. It looks to the past only for an empirical diagnosis of our practical participation, and not for any normative appraisal of that participation.Footnote 2 She does not say that we are responsible, now, for our past participation in structural harms, and she explicitly denies that key moral emotions are to be directed at those past participatory acts. For Young, it seems, we are never responsible for past contributions, and we are not to be subjected to the normal range of moral reactive attitudes for those contributions.

Now, several philosophers have criticized Young’s model, and because I think that their critiques miss the mark, I will spend the next few paragraphs discussing and responding to these critiques, showing that they fail to appreciate how Young’s model of responsibility-for-justice is indexical. Moreover, I will argue that they mistake a purely conceptual problem for a practical one.

To begin, Martha Nussbaum complains:

It seems to me that what we ought to say is that if A has responsibility R for social ill S, and she fails to take it up, then, when the relevant time passes, she is guilty of not having shouldered her responsibility. I think that this follows quite simply from the logic of ought: Young says that A ought to shoulder the burden; well, that appears to imply that if A doesn’t shoulder the burden A has done something wrong. (Nussbaum in Young 2003, p. xxi)

In the same spirit, Christian Barry and Luara Ferracioli attribute a basic flaw to Young’s system when they charge that ‘the assessments of behavior that this model will yield do not appear to be inter-temporally consistent’ (Barry and Ferracioli 2013, p. 255)

However, it is not entirely fair for Nussbaum to argue that the logic of our moral concepts dictates that Young’s model cannot be right. It is true that, ordinarily, we do not think that shirked ethical responsibility simply vanishes in a metaphysical puff of smoke, but this is only because we ordinarily cleave to the very conceptual scheme that Young is trying to complicate. Her proposal is that this can in fact happen, if we accept a supplementary, temporally indexical conception of responsibility that will operate alongside the more traditional notion. We can call this ResponsibilityFL, and we should allow Young to insist that this concept need not be constrained by the conceptual implications of the liability model. Similarly, we should say that in failing to appreciate that Young’s ResponsibilityFL is indexical, Barry and Ferracioli discover a ‘contradiction’ that simply is not there. There is no contradiction in saying that the sentence ‘tomorrow is Sunday’ can be true on Saturday and false on Sunday, because the concept ‘tomorrow’ is indexical.

However, conceptual schemes can be flawed in other ways. Rather than asking about mere coherence, we can ask, for example: is Young’s scheme one that human beings can use? The answer, I will now argue, is ‘no’. The real problem with the forward-looking approach is not that it is inconsistent or conceptually confused, rather, the problem is that it assumes that various key emotions and reactive attitudes can even in principle be excised from significant portions of human moral practice. In an ironic twist, the model emerges as a form of ideal theory: a conception of justice which can only be applied by creatures who are very different from us. Drawing on Charles Mill’s influential articulation, we can say that it is a model that assigns to human beings ‘completely unrealistic capacities’ (Mills 2005, pp. 167–168).Footnote 3 The reason we intuitively resist the notion that a person’s responsibility could just vanish into thin air is that we know, deep down, that we could not actually feel as though it had.

Structural Injustice and the Emotions

In what follows, I will rely on an assumption which should not be controversial: that human moral life is pervasively regulated by emotions, desires and conative states more generally. This is true at both the individual and the social level. At the individual level, emotional responses reflect our values and motivate us to act in accordance with those values. At the social level, I will assume that Peter Strawson was basically right in his description of our practices of holding one another responsible. The core message of his famous ‘Freedom and Resentment’ is that there is nothing more to the notion of moral responsibility than what is contained in the reactive attitudes we express to one another via such practices as praising and blaming. Once you understand that we are the kinds of creatures who regulate social behavior by feeling and expressing resentment, gratitude, love, pride, admiration, guilt, shame and so on, there is not much else you need to know in order to understand the concept of moral responsibility (Strawson 1974). This strong conceptual thesis (that there is nothing more to the concept) has been rejected by many, and I will not be taking it on board here. But as a matter of pure individual and social psychology, few deny that Strawson was right; even those who say that our concept of moral responsibility tracks some deeper metaphysical fact about us acknowledge that our actual practices of holding one another responsible are shot through with the emotional or conative content that Strawson identified. Moreover, and this is what matters for my purposes here, moral life in all known human cultures is pervasively regulated by backwards-looking emotional appraisals of behavior; for example, shame, guilt, pride and admiration.Footnote 4

These emotions, as I have already indicated, come in two varieties, the self-directed and the other-directed. Gabriele Taylor, in her influential work, calls the former ‘emotions of self-assessment’:

[I]n experiencing any one of these emotions the person concerned believes of herself that she has deviated from some norm and that in doing so she has altered her standing in the world. The self is the 'object' of these emotions, and what is believed amounts to an assessment of that self. (Taylor 1985, p. 1)

And, of course, she notes that emotions of social assessment concern the belief that another person has either violated a norm to which we are committed or failed to promote a value we take to be important (Taylor 1985, p. 21). So, my claim here is that backwards-looking emotional responses, of both the self-directed and other-directed variety, are core and ineliminable features of human moral psychology, and that as such, they play a leading role in motivating moral behavior. With this hopefully uncontroversial assumption in place, we can now ask: what would happen if we actually tried to deploy Young’s forward-looking, social connection model? What would happen if we tried to hold one another responsible for structural harms in accordance with its dictates?

Suppose we approach a few homeowners living in Sandy’s city and describe Sandy’s terrible, gut-wrenching situation to them. We inform them of their (fuzzy, complex) contribution to the plight of people like Sandy, but we try to explain that we do not mean to imply that they are to blame for Sandy’s plight. We insist, in accordance with the model, that we are not expressing resentment or anger at their being causally involved. We proceed to tell them that they merely have forward-looking ResponsibilityFL to join with other people in trying to ameliorate the overall situation, perhaps by symbolic acts of resistance against the economic system, perhaps by changing the system itself. Will the homeowners feel ashamed of their ongoing involvement in structural harms, even though we have explicitly tried to avoid provoking those feelings?

Remember, we are assuming that they are generally good people who do not aim at social inequality in any sense. That is, they accept our negative evaluation of extreme social inequality. Now, we might assume that they will not feel any backwards-looking moral emotions of the problematic sort, simply because people like this carry on all the time, untroubled by their own fuzzy contribution to large-scale social harms. But we should be more careful here. After all, as Young herself recognized, there are several popular methods for achieving this piece of mind. She lists:

  1. 1.

    Reification, the agent’s insistence that they ‘confront forces that give [them] no choice but to act as [they] do’ (Young 2003, p. 154),

  2. 2.

    Denying connection, the insistence on one’s own causal impotence with respect to the structural harm (Young 2003, p. 158),

  3. 3.

    Emphasizing Immediacy, stressing that one already has a great many duties and responsibilities to those one can directly interact with (Young 2003, p. 161) and

  4. 4.

    The ‘Not My Job’ defense, which simply involves the claim that others are better positioned to make relevant changes and therefore ought to do so (Young 2003, p. 165).

But we should ask: why do agents typically perform these defensive maneuvers? The answer is clear enough: for the same reason that anyone performs any such maneuver, namely, to suppress negative emotional responses. To banish them, if not from the mind entirely, then at least from immediate consciousness. In other words, such responses are defense mechanisms against negative self-directed moral emotions such as shame or guilt. This indicates that people like our good-hearted homeowners will almost certainly feel backwards-looking shame, guilt or regret, even if they immediately try to deflect or bury these feelings by insisting that, for example, their contribution to the problem is miniscule. Moreover, because these are basically good people, it is likely that this response will recur unbidden when the problem of structural harm is considered again, and that the response will need to be repressed each time it does. I conclude, then, that it is unrealistic to expect that the deployment of the social connection model will not provoke the very emotions it seeks to avoid or move past.

Does it matter that we insist to the homeowners that we are not blaming them? I do not think it does. Here is Young describing exactly how this is supposed to work:

We should not be blamed or found at fault for the injustice we contribute to, and we should not be blamed or found at fault for what we do to try and rectify injustice, even if we do not succeed … [H]owever, we can and should be criticized for not taking action, not taking enough action, taking ineffective action, or taking action that is counterproductive. (Young 2003, p. 144)

The difficulty here is simple. Socially and psychologically, there is no difference between blaming and criticizing. The typical speech-acts are the same, the social functions are the same, and most importantly, the characteristic emotional responses are the same. Imagine trying to comfort someone who has been subjected to a barrage of moral criticism by saying, ‘well, don’t worry, at least we’re not blaming you’. This will ring hollow, precisely because moral criticism is inexorably bound up with the kind of other-directed emotional responses that interested Strawson, such as resentment.

So, when we ‘criticize’ them in Young’s sense by pointing out their continued involvement in structural harm, or their failure to join with others in combatting the harm, our homeowners will almost certainly feel a negative self-assessment of some kind, and they will very likely interpret us as expressing some kind of negative emotional assessment in their direction. Nonetheless, this may not always be the case, and suppose we successfully move past or even somehow avoid this initial reaction, such that guilt, shame or regret have actually been precluded. Now, a new question will arise: what will provoke the homeowners into acting in accordance with the forward-looking duty we have just assigned to them? And how will we motivate continued action of the sort that is likely to be politically effective?

In cases of ordinary injustice, sorrow and compassion for the plight of people like Sandy will do the job. But can advocates for the social connection model allow our homeowners to experience or act on these emotions? It might seem as though they can, because compassion is merely a forward-looking emotion which normally fixes our gaze on what we can do (rather than on what we have done). But in fact this is not so simple, because forward-looking moral emotions are necessarily paired with backward-looking emotional responses.

Let me explain what I mean by that admittedly cryptic remark. When we hope for a result, we are disappointed when it does not come about. When we fear an outcome, we are relieved when it does not obtain. The deeper lesson is that emotions reflect our values, and since our valuations are normally stable, this guarantees that each forward-looking emotion will, when it reflects a valuation of a particular state of affairs, come along with a backward-looking emotion when that state of affairs comes about (or fails to come about).

Now, this claim—that forward-looking compassion and sorrow are necessarily linked to backwards-looking emotional appraisals—can seem like an empirical claim, one we need a great deal of evidence to believe. I am not so sure. After all, it may be that our judgments about who is genuinely compassionate are linked, conceptually, to our judgments about whether they feel retrospective guilt or shame in failing to aid. It would be entirely natural to say of someone who simply never experienced such retrospective emotions that they are not actually compassionate. Similarly, if I found that I never felt such retrospective emotions concerning my conduct towards someone, then I might wonder whether my allegedly ‘compassionate’ feelings for that person were not actually something murkier and less admirable. While individual deviations might be possible, it is not clear to me that the tight relation between forward-looking and backward-looking emotion is in fact an empirical one. It may simply be an element of so-called ‘folk psychology’, the background conception of mental states that allows us to pick them out and perceive their connections to action.

Yet, I understand that some may still hunger for empirical evidence here, and while I am unable to locate any direct tests of the hypothesis under discussion, I should mention two things. First, there is strong evidence that compassionate or empathetic people experience higher levels of guilt in general (Leith and Baumeister 1998; Hoffman 2000). This indicates that compassion leads to a heightened sensitivity to failure, which is exactly what you would expect if my hypothesis is correct. Moreover, a good illustration of this connection can be found in what is called ‘compassion fatigue’ (Potter et al. 2010; Figley 1995, 2002). In care-based careers such as nursing or psychotherapy, this refers to a kind of burnout, where a subject’s capacity to feel and be motivated by caring, compassionate feelings is dramatically decreased. Moreover, a fundamental cause of compassion fatigue is the slow realization that one can only sometimes restore a patient to full health, that one’s efforts to heal a patient will very often fall short of their goal. This is what Joana Duarte and José Pinto-Gouveia call ‘empathy-based guilt’, and after a large-scale empirical study they conclude that ‘nurses who are more prone to experience pathogenic empathy-based guilt… are particularly vulnerable to symptoms of burnout and compassion fatigue’. This, they emphasize, is due to the ‘negative self-oriented emotions’ elicited in particularly empathetic people by the continuing distress of others (Duarte and Pinto-Gouveia 2017). Moreover, many nurses are haunted by the thought that the institutional constraints they are under have caused them to fail to even try properly, that is, they worry that they ‘care’ for patients in a desultory and ineffective fashion. All of this leads to compassion fatigue. But now, notice that this empirical result implies that retrospective guilt, shame and regret are tightly connected to compassion. To work compassionately towards the health of the most vulnerable patients is psychologically risky, and a fundamental source of that risk is the potential for serious negative self-assessment, for the unshakable sense that one could have done more.

Thus, even when compassionate people do make strong efforts to aid others, their inability to secure an optimal outcome haunts them. Surely, then, we should expect that compassionate people who do not even make the effort will be even more prone to feelings of regret, guilt or shame. Thus, to return to the case at hand, for our homeowners to feel forward-looking compassion for those they can help is also for them to feel guilt, shame or regret if they are confronted with their own failure to join others in helping. This is because in order to be motivated to change at all they must strongly value the state of affairs in which they fight structural harms, and so they must regret (to some extent) failed opportunities to do so.Footnote 5 Yet, guilt and shame are precisely what our homeowners are not supposed to feel if they fail to act. So, the social connection model cannot encourage them to feel sorrow or compassion for people like Sandy, because those emotions will almost inevitably encourage the sort of backward-looking responses we are not supposed to be feeling or expressing. Yet, this seems to imply that sorrow and compassion for Sandy’s plight should be firmly discouraged, because such feelings will almost invariably provoke backward-looking responses. This, indeed, is how most of us relate to people like Sandy; we simply ignore her in order to avoid compassion and sorrow altogether, not wanting to deal with guilt or regret later on. But this, of course, is exactly why we do nothing. But now, one wants to know, what exactly is going to motivate the homeowners to act on their forward-looking responsibility?

Now, it might be thought that other emotions can do the job. One emotion that is prominently linked to perceptions of injustice is anger. As Aristotle rightly says, ‘it is apparent injustice that arouses anger’.Footnote 6 Can’t we simply feel forward-looking anger at a social or economic system without also feeling negative emotions of self-assessment when we fail to act? With my basic analysis now familiar, we can see that this is unlikely to work out. I may experience anger at injustice, anger at a system that fails so many, and may thereby resolve to do something about it by joining with others who are fighting the system. Putting aside one obvious problem, which is that this response involves emotions of compassion and sympathy towards those being failed, how can I later confront my own failure to do anything at all without feeling any negative self-appraisal? After all, if I am truly angry at the system, have I not, through my own inaction, reinforced my own role in that very system, complicit in its evils? Have I not reinforced my social connection to injustice in precisely the sense that Young outlines? Once again, anger, as an emotion, is grounded in a certain evaluation: the system, I think, is deeply unfair or unjust. If that evaluation remains stable over time, I must feel some sort of regret or disappointment at my own failure to take action. The evaluation that gives rise to the anger also gives rise to a negative self-appraisal.

We should also remember that while my focus has been mainly on the subject’s own self-assessments, the social connection model is also meant to preclude our expressing disappointment or blame towards others who also fail to take action. And here, I simply cannot see how I am to be angry at a system and not feel precisely those sorts of negative emotions towards people around me who remain inactive or complicit.

The problem only becomes worse when we consider something that Young explicitly admits, namely, that positive backward-looking appraisals are also strongly discouraged by the social connection model. For suppose we solve the problem of moral motivation, we may still ask: what if our homeowners do successfully join together with others to combat structural injustice, such that a great many people like Sandy have access to affordable housing? Does the social connection model allow them to feel or express pride in their accomplishment? Does it permit third parties to express love or admiration for them? It seems not: these would be precisely the sorts of expressive acts that the forward-looking model instructs us to avoid. As Young says, the point of the model ‘is not to praise or blame’ (Young 2003, p. 109, italics mine), and I take it that the idea here is that model will become dysfunctional if people start to praise and blame one another in familiar ways. That is, it will simply collapse into the liability model. But this, I emphasize, is an extremely unwelcome result, because it is unclear that the actual social change Young wishes to provoke could possibly occur without these emotional expressions.

It is important to remember that in developing a temporally indexed conception of responsibility for structural injustice, Young has made it such that there is quite literally nothing for an effective opponent of structural injustice to feel proud of. If this is denied—if Young were to say that there is some successful exercise of agency about which an agent ought to feel pride—Nussbaum’s criticism would return with a vengeance, since it would be difficult to see how the model is forward-looking in any real sense.

Now, Young could try to incorporate admiration and pride into her model. She might just say that her temporally indexed conception of responsibility only applies to moral failures and not to moral successes. But this is an extremely unstable thought. How is an agent supposed to be constituted so as to feel pride at successfully performing an action while at the same time being such as to not feel any sort of guilt, shame or regret for not performing the same action? I can only repeat that emotions reflect our values, and that for this reason these various responses are deeply intertwined in human psychology.

We can now see why the model faces severe practical difficulties. For how are populations of human beings supposed to band together to form effective coalitions against structural injustice if the only known mechanism for banding them together—a key cluster of emotions and reactive attitudes—has been almost entirely removed from the picture? Creatures very different from us might be able to use the model, to keep their moral gaze firmly fixed on the future. But that cannot be of interest to a philosopher who wishes to rectify injustice as it appears in the human world.

Of course, I should stress that people in fact do band together to fight structural injustice, and those who read Young’s pioneering work are far more likely to do so. My point is just that this political inspiration will have its roots in the very emotionally laden responses that Young seeks to avoid. The model can only inspire change in agents who are deeply self-deceived about what they are doing.

These observations can lead us to see what is wrong with any interpretation of the liability model as a ‘liberal’ model, or with Zheng’s claim that addressing structural injustice requires a ‘post-liberal’ turn away from metaphysics (Zheng 2018). The forces which give rise to and sustain the liability model are much older and deeper than any particular philosophical theory or legal practice. Neither Locke nor Kant brought them into being, nor can they be dispelled by the recognition that it might be better in some vaguely consequentialist sense if they were dispelled. There are no known cultures without analogues of guilt, shame or other backwards-looking reflections of forward-looking emotions, nor are there moral cultures without something like moral pride or emotionally laden admiration of past deeds. Here is Homer’s Achilles, crying out in shame at the death of his friend Patroclus:

He died far from home and he needed me to protect him. But now, since I’m not going home, and wasn’t A light for Patroclus or any of the rest Of my friends who have been beaten by Hector, But just squatted in my ships, a dead weight on the earth, I stand alone in the whole Greek army (Illiad 105–110).

Here, Achilles is holding himself accountable, in accordance with the liability model, when thinking of the deaths of his friends. Yet, this is quite obviously not because he inhabits a liberal moral culture or because he is caught up in any distinctively metaphysical project.

Though we should be wary of appeals to human nature which are merely ideological smokescreens for injustice, it is nonetheless difficult to see how these various emotional responses could be anything other than an ineliminable feature of our sense of personal responsibility, or how we could fight structural harms without them. For this reason, any system which attempts to assign duties, reasons and oughts to individuals in abstraction from those feelings is doomed to practical irrelevance. Our task, going forward, is to develop a theory of structural injustice that respects the critical role played by the moral emotions in human social life.

Looking Forward

What might such a model look like? I lack the space to fully articulate an alternative here. However, I am convinced that the concept of responsibility is creating enormous trouble for theorists in this area. Once we take up this concept as our primary theoretical target, we are led directly to a confrontation with the social and psychological forces that sustain the liability model. We may pretend to be excising that model from our theory and practice, yet, as I have shown, it will continue to haunt us in various ways. Yet, this model, as Young points out, is of dubious help in motivating real structural change.

But why make responsibility our key theoretical notion? After all, one alternative is to focus on the distinct concept of a reason for action. That is, instead of asking whether our homeowners are responsible for structural harms, why not simply ask whether they might not have powerful practical reasons to fight those harms? We often have powerful practical reasons to remedy certain situations while remaining entirely free of personal responsibility for the situations themselves. To adopt an example from Robert Goodin, the person who dropped the child from the window is not the person below who can catch the child (Goodin 1995). Only the former person bears responsibility for the situation, but the latter person can easily have powerful practical reasons to remedy it.

What are the sources of these reasons? Meta-ethicists continue to fight this one out, and we need not take any position in the seemingly interminable internalism/externalism wars here. All we need to think is that sometimes an agent’s emotions will explain their practical reasons.Footnote 7 For example, in the above case, at least some of the powerful reasons that most of us will have for catching the falling child are grounded in the extraordinary shame and/or guilt we will feel if we stand aside and do nothing. This, I must emphasize, leaves it entirely open whether there are not so-called external reasons to catch the child, reasons that apply to any agent no matter what their emotional configuration.

So, returning to our homeowners and to the plight of people like Sandy, we can say that their sense of unease and shame at being involved in structural harm can help to generate reasons for action. It is true that they, themselves, are nearly powerless to change the larger system, but as Young rightly says, these reasons might simply be reasons to join or to try to organize with others, reasons that are contingent on the availability of a collective that can be joined or organized.

Alternatively, we can say that these are integrity-based reasons, grounded in an emotionally laden refusal to be part of an unjust system, no matter how insignificant their role happens to be (Boey 2016; Halfon 1989). ‘This might happen, but it will not happen through me’, is a perfectly familiar ethical thought, one made famously salient in Bernard Williams’s famous case of Jim, who is told by a warlord that if he does not shoot one innocent person, the warlord will shoot 20. Jim’s integrity-based reasons, in this case, derive from his personal inability to shoot anyone, regardless of what the consequences will be (Smart and Williams 1973). Whatever the right thing to do is in such a case, it is clear to many of us that the answer cannot simply be reduced to the question of what difference Jim can make in some consequentialist sense. Or, if that is the answer, it is not one that can be expected to make sense to ordinary human agents with ordinary moral emotions. Rather, an answer that will make sense to Jim must take his personal integrity into account by giving weight to his own commitment to non-violence. Similarly, in the case of structural injustice, we can suggest that many people will have strong integrity-based reasons to reduce their social connection to structural harms, even if such a reduction makes no appreciable difference.Footnote 8

Indeed, my sense is that integrity will be the most powerful notion here, one that actually explains how so many activists and citizens are moved to try to change their behavior when their individual contribution makes little to no difference. They simply cannot stomach the idea of being a cog in such a terrible machine, and the fact that they are an insignificant, eminently replaceable cog is beside the point. This sense of integrity is, I submit, an emotional response which can supply powerful practical reasons to fight structural injustice and which can motivate us to act in accordance with those reasons.

All of this being said, regardless of whether we conceive of these as reasons to make a difference by joining with others or as reasons to remain true to one’s own personal integrity, the key point for my purposes is that this reasons-based model would not seek to prohibit or avoid backward-looking emotional responses. Indeed, many of those responses would form the backbone of our reasons to pursue a more just world. To sense that one cannot live honestly with oneself if one continues to support unjust systems is to fear the familiar sting of guilt and shame, and a reasons-based model can more than happily encourage and accommodate such backward-looking emotional appraisals.

Many difficulties remain for such an approach. In particular, it will be unclear just who has these reasons and what their relative strength will be. Moreover, we will have to ascertain what sorts of moralized responses are appropriately directed at agents who act (or fail to act) on such reasons. And of course, such reasons will probably not, on their own, motivate a full-scale remedy for problem of structural harm. However, this conception of the ordinary individual’s role is perfectly compatible with further (entirely sensible) ideas about the obligations that governmental and corporate agents may have. In any case, the key point is that these reasons, unlike the obligations that are said to arise from Young’s ResponsibilityFL relation, stand a fighting chance of motivating us to improve the world. And that, at least, is a start.