Abstract
This article argues for a systemic conception of freedom as non-domination. It does so by engaging with the debate on the so-called coalition problem. The coalition problem arises because non-domination holds that groups can be agents of (dominating) power, while also insisting that freedom be robust. Consequently, it seems to entail that everyone is in a constant state of domination at the hands of potential groups. However, the problem can be dissolved by rejecting a ‘strict possibility’ standard for interpreting non-domination’s robustness requirement. Frank Lovett and Philip Pettit propose to restrict the relevant domain of possible worlds by reference to two epistemic conditions pertaining to potential group members. I argue that this strategy unduly limits non-domination’s critical potential. I then argue that a suitably systemic conception of domination avoids this problem. By placing explanatory emphasis on social institutions, and how these bear on the feasibility of individual and collective action, a systemic conception of non-domination avoids the coalition problem in a way that retains its critical potential. The article clarifies the relationship between the rule of law and the social norms and objects to the claim that non-domination is bound to deem the latter irrelevant from the point of view of freedom.
Introduction
‘Neo-republican’ freedom as non-domination promises to hold explanatory power in cases where ‘liberal’ freedom as non-interference is thought to struggle. According to its proponents, it can unambiguously diagnose workers without sufficient labour protections, women in patriarchal societies, disenfranchised minorities and – the classic example – slaves owned by benevolent masters, as unfree. Members of these groups are dominated because others have power to interfere in their acting, at will and without reliably expecting sanctions. Thus, they lack freedom even if their dominators never actually interfere. Agents can suffer domination at the hands of individuals who possess power by virtue of their group membership or by groups themselves when their members have the power to act in combination. This seems to make non-domination especially well-suited to ground social critique, apt to capture the freedom-constraining nature of racist, patriarchal and material oppression. To be free from domination is to enjoy robust non-interference from individuals or groups.
Perhaps ironically, it is also this insistence that social freedom must be sensitive to the power held by groups that gives rise to non-domination’s most serious challenge. According to the so-called ‘coalition problem’, freedom as non-domination is self-defeating, because it demands protection from powers that simply cannot be controlled: since no group of people is so powerful that no other could possibly interfere with its acting, everyone is constantly dominated by the existence of such potential groups (Dowding, 2011; Simpson, 2017). All potential groups hold such dominating power, regardless of the likelihood of their formation. Thus, it is not only the oppressed who suffer domination at the hands of their oppressors; the oppressors suffer domination at the hands of potential groups consisting of the oppressed. This result appears bizarre. But – so the objection goes – it cannot be avoided without sacrificing non-domination’s coherence or diagnostic power.
In this article, I aim to defend freedom as non-domination from the coalition problem. Further, I aim to show that only a conception of domination that is suitably systemic can do so in a way that retains non-domination’s critical potential. I will therefore not only be concerned with dissolving the coalition problem but also be concerned with pointing out the shortcomings of a recent response to it, offered by Frank Lovett and Philip Pettit (2019), the two most prominent proponents of freedom as non-domination. In short, the coalition problem is generated through a maximally wide interpretation of non-domination’s robustness requirement. Lovett and Pettit’s strategy is to narrow the domain of possible worlds captured by the robustness requirement by introducing two epistemic criteria pertaining to potential group members. This dissolves the coalition problem but, I argue, does so in a way that excludes cases where we want to keep the verdict that the agent in question is dominated. I then spell out my own strategy, arguing that the relevant domain of possible worlds ought to be demarcated by analyses of existing social institutions.
My argument is structured in the following way. In the first section, I introduce freedom as non-domination and the coalition problem. I clarify the specific challenge the problem poses. In the second section, I discuss Lovett and Pettit’s response, arguing that it limits non-domination’s critical potential. The third section begins my positive proposal by outlining Cécile Laborde’s distinction between agent-relative and systemic domination. I argue that the latter is more important for analyses of freedom and that it requires placing explanatory emphasis on social institutions. The fourth section elaborates on this claim by showing how the presence of social institutions is consequential for the feasibility of individual and collective action. Dominating power, I argue, is constrained by feasibility. The fifth section extends this point by demonstrating how informal social institutions enable and constrain abilities to bring about states of affairs in ways that undermine the coalition problem. Drawing on the literature on epistemic injustice, I object to the claim that non-domination is bound to deem social norms irrelevant from the point of view of freedom. To the contrary, I argue that the rule of law must be supported by egalitarian social norms to bring about robust non-interference. Analyses of social freedom ought, therefore, to be centrally concerned with the social norms and conventions that structure allocations of credibility and status. The sixth section concludes this article.
Non-domination, robustness and domain-restriction
Who is free to act in the social world, and why? On the standard account, an agent is free when and because they do not face constraints imposed by the agency of others. An increasingly popular alternative to this ‘liberal’ account of freedom as non-interference is ‘neo-republican’ freedom as non-domination (Lovett, 2010; Pettit, 1997, 2012; Skinner, 1998). On this view, an agent is free to the extent that they are not dominated, which is the condition they suffer when others have the power to introduce constraints on their acting, according to their own will and without reliably expecting sanctions. 1 Freedom from domination requires robustness. When the non-interference you currently enjoy remains constant across a range of possible worlds, you are free. When there are nearby possible worlds where your action is constrained, and you have no means of control over the proceedings that could bring those worlds into being, then you are dominated. 2 In that case, your unconstrained acting can be explained by reference to another’s decision not to interfere: you are dependent on their will. Bringing about non-domination is thus a matter of constraining abilities to interfere by introducing rules, procedures and goals that work as external checks on individual wills (Lovett, 2010: 96; Pettit, 2012: 63). For this reason, freedom as non-domination is definitionally tied to the rule of law (List, 2006). An order of law enforced by the state or its functional equivalent revokes the possibility of interfering with others without facing material sanctions. The law functions as an external constraint on individual wills and is, unlike those wills, not susceptible to change.
Proponents of freedom as non-domination hold that collectives with a unified will, as well as individuals, are agents of power. An agent may thus be free from domination at the hands of any one individual but still be dominated by a gang, or a corporation, or a state. This is an attractive feature of the theory, because it can account for an important aspect of our experience of agency, namely, that groups can exercise power in ways that constrain our freedom. Further, this gives the theory a distinctively critical edge: by accounting for imbalances of power between social groups, non-domination seems well equipped to theorise loss in freedom stemming from racist, patriarchal or material oppression (Costa, 2019; Laborde, 2008). 3
In conjunction with the robustness requirement, however, this focus on social groups seems to generate a problem. If I could be constrained in my acting by a (currently unorganised) collection of individuals acting together, then I am seemingly dominated by that possible coalition. And if this is true, then I will always be dominated, since ‘for any individual i at any given time, there is always a coalition of others {j, k, l} who could stop i from doing x’ (Dowding, 2011: 310). This is, as Keith Dowding (2011) calls it, the coalition problem.
It is clearly a problem for a theory of social freedom that it systematically delivers the result that everyone is unfree. Anyone who is not a sceptic about social freedom will thus feel the pull of the coalition problem. It looks the problem can easily be dissolved by putting pressure on the thought that just any random collection of individuals has the power to act as a group. Pettit recognised the coalition problem in his earliest expositions of freedom as non-domination (1996: 580; 1997: 54–55). In these writings, he responded to the problem by drawing a distinction between actual and potential abilities to interfere and arguing that only those individuals who had already constituted themselves as a coherent agent could ‘actually’ dominate. This response is weak. The robustness requirement makes the appeal to groups that have actually formed arbitrary, since there are possible worlds (maybe even nearby) in which the potential members do form the group and interfere. Therefore, at least some potential groups must count as holding dominating power.
A more promising response works by reference to the necessary conditions for group agency. While a group of individuals could have interfered with my acting if they formed a coherent agent, they will not always be able to do so. Establishing group agency is demanding. It requires that the individuals who constitute the agent are capable of meeting certain standards of rationality as a collective (List and Pettit, 2011). Once incorporated, they must be capable of displaying stable and consistent attitudes in their interpretation of, and acting in, the world. Not all collections of individuals can meet these standards and thus not all collections of individuals hold dominating power by way of their potential for acting together.
This response seems plausible, but it raises a new problem for the non-domination conception. Holding potential group agency as the relevant standard for determining which collections of individuals hold power as a group seems incompatible with another key commitment of the theory: the rule of law requirement. Proponents of freedom as non-domination hold that the state or its functional equivalent is necessary to avoid dominium, that is, relations of domination between citizens. The state requires vast powers to carry out this mandate, and it must be organised so that it can stably and consistently deploy these powers. But when the state is organised in this way, it meets the standard for group agency (List and Pettit, 2011: 40). Thus, we now need a mechanism for avoiding imperium, that is, domination of the citizenry by their state. This is not just a matter of institutional design. Since the rule of law requirement is built into the definition of non-domination (List, 2006: 211), this is a concern for the theory as a whole.
Pettit’s influential argument is that the citizenry avoids imperium when it holds the power to control the state by exercising popular control (2012: 160). The people enjoy popular control when they have the ability to contest the state’s decisions and, ultimately, to interfere with its actions. This control is secured when the citizenry possesses a resistive character that enables organisation in the face of an unresponsive or oppressive state (Pettit, 2012: 174). The problem with this response, as Thomas Simpson (2017) argues, is that ‘the people’ cannot plausibly be thought to qualify as a potential group agent. This shows that there must be room in non-domination’s social ontology for a category of collective agency that is less cohesive and demanding than full group agency. Simpson calls this category ‘teams’: Teams have a degree of coordination that makes their action distinct from that which emerges from individuals interacting without any joint intention. But they do not have the degree of coordination that group agents require. (2017: 40)
To illustrate this claim, Simpson offers two pairs of examples. I will rely on these throughout. It is therefore worth spelling them out in full. 4 The first pair is variations of one of the motivating cases for introducing a robustness requirement on freedom in the first place: the master and the slave. In Simpson’s version, there is one slave and three masters. None of the masters are ‘strong enough to alone interfere with the Slave’ but each of them can successfully do so with the help of one of the other (Simpson, 2017: 32). In the case where one master wants to interfere, the slave is dominated by both of the others, since either of them can bring about the interference by changing their will. But we also want to say that the slave is dominated in the case where none of the masters want to interfere. If this is the case, it is their potential to act in combination that is the source of the slave’s domination. 5
The second pair of examples invokes the Indian caste system, another social institution that stirs deep intuitions about (un)freedom. In the first variation of the example, a Dalit (‘Untouchable’) is breaking a caste norm, taking water from a communal well that is also used by upper castes, including the Brahmins: Suppose the Dalit is a physically strong young man, carrying a knife as self-protection, well able to defend himself against a lone attacker. While pumping water, the Dalit notices an angry Brahmin looking to incite someone so that they can together attack him. (Simpson, 2017: 33) Now imagine that each villager has come to repudiate the caste system as a tool of oppression, and is convinced of the dignity of each person regardless of birth. Caste norms have no social power. The Dalit now pumps water without even thinking of possible threats. His situation is akin to the slave in [the case where no master wants to interfere]. The only reason for supposing that the Dalit is not dominated is the internal change in the villagers’ moral conviction. (…). The villagers (…) could repudiate their convictions and attack as easily (or with as much difficulty) as the Master could repudiate her benevolence and interfere [with the slave]. Nothing else restrains the villagers from attacking. (2017: 33–34)
As the specification of the Dalit’s knife makes perfectly clear, the conception of power on display here is one of the physical force. Neither the masters nor the Brahmin are strong enough to interfere individually, but they would be by combining their strength. There are possible worlds in which they do so, and this suffices for the slave and Dalit’s domination. But there are also possible worlds in which teams interfere with the masters and the Brahmin, and this must also suffice for their domination. As Simpson writes: ‘it does not matter how improbable it is that they will do so; the possibility that they may makes [any agent] unfree’ (2017: 34). This exposes the core of the coalition problem. It arises when we interpret non-domination’s robustness requirement according to a ‘strict possibility’ standard on which all possible interferences count as directly relevant for social freedom (Dowding, 2011: 310; Goodin and Jackson, 2007; Simpson, 2017: 34).
Why should we think that the robustness requirement ought to be interpreted according to such a standard? Conceptually, there is nothing that demands this (List and Valentini, 2016: 1048; Pettit, 2008). Thus, if we could specify the domain of possible worlds that is to count as directly relevant for assessments of social freedom in a way that excludes the most improbable ‘team formations’, the coalition problem would be dissolved. The objection, then, must be that there is no principled way of doing this that delivers plausible results with regard to which collections of individuals actually hold power as a potential team (cf. List, 2004: 77). One response to this challenge is to argue that we specify the relevant domain by reference to substantive moral values. For example, we could argue that only those potential teams whose power lacks moral justifications count as dominating. However, many authors – critics and supporters of non-domination alike – think that moralising the definition of freedom reduces claims about freedom to claims about morality which, in turn, deprives us of a crucial concept for when we engage in moral evaluation (Carter, 1999: 70–71; cf. List and Valentini, 2016: 1059). Indeed, Dowding’s argument is precisely that the coalition problem exposes how non-domination is inevitably moralised (2011: esp. 305–308).
Now, some proponents of non-domination reject that it is problematic to advocate a moralised conception of freedom (Costa, 2019; Laborde, 2010). But, as I will show, the coalition problem does not provide a new locus for this internal disagreement. We can restrict the domain of relevant possible worlds by reference to a non-moralised account of systemic power. To set the stage for this argument, it will be instructive to see the shortcomings of a different strategy.
Domain restriction by awareness and strategy
Recognising the force of the challenge posed by the coalition problem, Lovett and Pettit (2019) have recently proposed a strategy for the required domain restriction. This strategy, however, is agent-relative in a way that generates a different problem. As I will argue, it blunts non-domination’s critical capacity. The systemic alternative I sketch in the following sections avoids this problem and is, I argue, therefore superior.
As is required to respond to the coalition problem, Lovett and Pettit object to the claim that the mere possibility of team formation suffices for dominating power. In addition to possible will convergence, they argue, potential team members must satisfy a condition of awareness: each must know that their wills have converged (Lovett and Pettit, 2019: 376). Further, they must also satisfy a strategy condition: there must be a salient strategy of interference available to the team members as a matter of common awareness (Lovett and Pettit, 2019: 376). ‘Under normal circumstances’, Lovett and Pettit argue, these conditions will not be satisfied (2019: 377). Therefore, we cannot generally expect that individuals will be willing to risk engaging in an interference with the hope that others will join. The existence of possible worlds with successful interferences by spontaneous teams does not compromise freedom. The relevant domain of possible worlds is restricted to include only those team interferences where the awareness and strategy conditions were satisfied prior to the actual interference.
Thus, on Lovett and Pettit’s account, the relevant domain of possible worlds is demarcated by knowledge possessed by individuals in the actual world. This dissolves the coalition problem, since now only a subset of collectives – including ‘the people’ in some favourable circumstances (Lovett and Pettit, 2019: 382) – hold power as potential teams. An important caveat to this is that possible interferences falling outside the relevant domain are not completely irrelevant for freedom. Since freedom as non-domination is supposed to ground a republican political theory concerned with the design of social institutions, such possible interferences matter insofar as different institutions make them more or less probable (Lovett and Pettit, 2019: 365–366). However, this probability does not count ‘directly’ for current assessments of freedom.
Although the introduction of awareness and strategy clearly rebuts the coalition problem, it comes at high cost. This becomes apparent already in Lovett and Pettit’s own discussion. Here, they are led to argue that Simpson is wrong to treat the unfreedom of slaves as axiomatic, on the non-domination view: (…) for all that Simpson’s descriptions of the scenario imply, the awareness and strategy conditions might not be satisfied. If they are not, the masters will not have the capacity to interfere, and thus not constitute a capable team holding sway over the slave. Thus, it is false to claim, as he does, that necessarily the [masters in his examples] dominate the slave. (Lovett and Pettit, 2019: 378)
This is a striking concession. The fact that the slave is a slave – that someone owns him and can decide what he does – is now only derivatively relevant for assessments of his freedom. This opens for the unappealing result that a highly regulated regime of slavery that prohibited masters from owning slaves individually could be compatible with the social freedom of those slaves with a benevolent group of masters, so long as interference was highly improbable. But this is precisely the kind of consideration proponents of non-domination take issue with. 6 As Lovett and Pettit write elsewhere, willingness to consider probabilities or preferences as relevant for the freedom of a slave ‘does not square with our strong sense that slavery, whether one’s master is good, bad, or indifferent, is the quintessential state of unfreedom’ (2009: 14). The introduction of more masters does little to change this sense.
Lovett and Pettit’s strategy raises further concerns about non-domination’s critical potential. First, their presupposition that subjects will fail the eyeball test because they know that they might become subject to ‘actual’ domination poses a question about those who are simply unaware of the power held by their ‘potential dominators’. If a colonial subject’s rulers do not fulfil the awareness and/or strategy condition, and he is unaware that they count him as their subject, does that make him free from domination? To respond that he has reason to fail the eyeball test, and that he therefore counts as dominated, seems both to go against the test’s focus on experience (Pettit, 2012: 84–87) and to add a moralised component to the definition of domination. Second, the explicit reference to knowledge on the part of dominators reinforces an existing concern about social practices that are sustained at the unintentional level. As Sharon Krause (2013) argues, insofar as domination is cashed out in terms of intentional capacity for interference, it cannot diagnose obstacles to agency that stem from widespread prejudice. Widespread prejudice has the effect of making individuals in positions of authority think that they exercise their powers in fair and accurate ways when they, in fact, systematically disfavour certain groups. Needless to say, this problem is pervasive, and it is problematic if non-domination cannot deliver the verdict that persons are made less free by such prejudice. 7
In short, Lovett and Pettit’s strategy for dissolving the coalition problem refers to resources – in this case, epistemic – held by individuals. In so doing, they constrain the scope of the robustness requirement in a way that renders non-domination vulnerable to a critique of insufficiency (cf. Markell, 2008). Thus, the coalition problem seems to have exposed that the ideal of non-domination is either demanding to the extent of impossibility or so limited in its critical ability that it loses its original appeal. Thankfully, both of these results can be avoided. In the next section, I begin my argument that the scope of robustness should be constrained with reference to social-systemic, and not individual, resources.
Domination: Agent-relative and systemic
Recall that Simpson diagnoses the slave’s domination by reference to the possibility of his masters combining their individual physical resources. Lovett and Pettit accept this basic premise but add two epistemic conditions pertaining to each individual master. To get around the problem that this delivers the result that the slave is not necessarily dominated, they argue for an indirect strategy of diagnosing his domination.
Consider now a different strategy for explaining the slave’s unfreedom. Drawing a distinction between agent-relative and systemic domination, Cécile Laborde argues that it is the latter that ought to be the central object of conceptual and normative analysis (2008: 152–155; 2010: 56–58). While it is clear that individuals and groups exercise power over others by virtue of their superior strength, the more interesting instances of power are those derived from a superior location in social structures assigning roles, rights and credibility. 8 It is this power that renders some robustly vulnerable to the will of others. When systemic features of society enable domination, those at the receiving end of it are unfree, but not simply in virtue of the resources of an individual or group. Instead, their domination depends on a set of beliefs, norms, and practices that is upheld by a sufficiently large portion of the relevant society. On this view, the masters do not dominate the slave primarily through their conjoint capacity for exercising physical force. Their dominating power has its source in ‘systemic features of slavery as a deeply entrenched, institutionalized set of rules and conventions’; it is the ‘institution of slavery that [provides] the resources and background structure for the domination of slaves by their masters’ (Laborde, 2010: 57). In other words, that there is an institution of slavery is more relevant to our diagnosis of the slave’s domination than is the observation that his masters can form a team to interfere with his acting. On a systemic view of domination, analyses of social freedom ought to be centrally concerned with the social structures – created and sustained by us – that enable people to exercise power over others. 9
The point that domination is primarily systemic is important, because it highlights a fact that is easily distorted by the coalition problem; people depend for their power on social conditions that are sustained by other people. The masters’ domination of the slave cannot be explained without reference to the fact that there exists an institution of slavery, that is, a system of property that recognises human beings as appropriate objects of ownership. For that system to exist, people must recognise it and act in accordance with its demands; it must structure their agency. As Iris Marion Young explains, when people act, they are trying to do two things at once: (1) They are trying to bring about a state of affairs that they intend, and (2) they are reproducing the structural properties, the positional relations of rules and resources on which they draw for these actions. (2011: 60)
Social institutions both enable and constrain. They make easier the achievement of the actions and outcomes available within a relevant institution, but they make it more difficult to achieve those states of affairs, or to act in ways, that are prohibited by that institution. The more deeply entrenched an institution is, the more difficult it will be to act contrary to it, simply because others will fail to treat the action as a legitimate way of achieving the desired end. In the case of social institutions that, like slavery, have also become legal institutions, such violating actions will face legal sanction. This, unsurprisingly, matters for social power.
Take the case at hand. The source of the masters’ power is that they can assert rights of ownership over the slave with help of their social surroundings. If we imagined Simpson’s example taking place in a setting with no recognisable system of property beyond what the masters enforced through their combined (agent-relative) power, it is uncontroversial that the non-domination conception would deliver the result that also the masters are dominated. This is because the absence of formal institutions prescribing rules against interference would render the masters vulnerable to potential teams along the lines highlighted by proponents of the coalition problem. Indeed, this is the reason why freedom as non-domination includes the rule of law requirement: since no individual holding of agent-relative power (‘strength’) can ever be sufficient for guaranteeing robust non-interference at the hands of others, freedom from such interference requires the introduction of systemic constraints on power in the form of the state. But when such systemic constraints on power are unequally allocated, they solidify the power some hold over others. This is the main source of the domination of slave, not the combined physical force of the three people who call themselves his masters.
It might be unclear where this argument diverges from Lovett and Pettit’s. Do not social institutions enable domination precisely when and because they ensure that the awareness and strategy conditions are satisfied? It is true that social institutions can work to ensure that awareness and strategy are satisfied (cf. Lovett and Pettit, 2019: 380). However, they can enable domination more broadly, too. As I will argue, although awareness and strategy are jointly sufficient for domination by potential teams, they are not individually necessary.
Social institutions and the feasibility of acting
In this section, I propose that the concept of feasibility is useful for explanations of who hold social power and why. Only those who can feasibly bring about given states of affairs have the power to do so. Social institutions, as I started explaining in the previous section, have a key explanatory role in assessments of feasibility.
The claim that feasibility is a constraint on social power is plausible for the simple reason that we typically do not think that people have the power to do things they can only bring about accidentally. This does not mean that only easy actions are feasible. For example, I may find it exceedingly difficult to speak in front of a particular crowd of people. Yet, this does not show that I am unable to do it; it only shows that it is improbable that I do it (Estlund, 2014: 119). Sufficiently incentivised – someone may convince me that it is urgent that I do so – I will do it. Following Zofia Stemplowska (2016), I thus understand feasibility to track whether an agent could bring about an action if properly motivated. This means that also feasibility is a modal concept: it is feasible for an agent to act when the realisation of that action by that agent is robust ‘to variation (within a salient range) of the initial micro-conditions of the relevant circumstances’ (Southwood and Wiens, 2016: 3048). On this model, changes in individual behaviour represent the micro-conditions, and the relevant circumstances are the (currently existing) systemic features of society. The main claim of freedom as non-domination can now be stated as follows: that someone could receive an incentive that makes them choose to interfere with your actions, and not face any form of sanction for so doing, is a fact that bears on your social freedom. 10
One immediate problem with this account is that it seems to be open to the possibility of truly benevolent slave masters, whose normative commitments will simply make them unresponsive to incentives, thereby making interference unfeasible for them. 11 There are three things to say about this. First, as Stemplowska argues, being unresponsive to incentives because of one’s normative commitments does not entail genuine motivational inability precisely because (not) acting from normative convictions ‘expresses our agency rather than limits it’ (Stemplowska, 2016: 280-281, note 21). Second, and relatedly, it is important to remember that the threshold for interference here is not set at drastic actions like torture or murder. It suffices for the slave’s unfreedom that there exists an incentive that will make his master willing to restrict their action in any way. Third, and most importantly, what we are motivationally capable of doing to other people is highly dependent on social institutions themselves. As Franz Fanon (2001 [1961]) powerfully argued, legal structures that grant a privileged class vast powers over others have deep psychological effects on both the oppressors and the oppressed. This, by itself, gives us strong reason to doubt that the existence of truly benevolent masters is psychologically realistic. It should be an uncontroversial historical statement that slave societies are bound up with ideas of moral and epistemic superiority; it straightforwardly explains why violent interference has typically not been outside the scope of slave masters’ motivational abilities. 12
The above-mentioned conception of feasibility can be deployed to rebut the coalition problem. A condition on feasible action is knowledge, or at least reasonable certainty, of how to bring the action about. In the individual case, this is obvious. It is not feasible for me to play a piano concert, because I do not know how to, and no incentive can change this (cf. Pettit, 1996: 580). 13 Similarly, in the case of collective action, this epistemic condition entails that we cannot just look at the motivation of each member. We must also consider the resources each member has for working out what they have to do, how to do it and in what order their individual actions are to be sequenced. When people are in situations where they have no established ‘method of deciding, communicating, and coordinating together’ (Stemplowska, 2016: 283), the individual motivation of each to carry out the collective action is insufficient for the feasibility of their bringing it about. For illegal actions, the rule of law is supposed to function so as to block the possibility of establishing such a method: attempts to instigate attacks on individuals or groups will be prosecuted. Thus, although each may be motivated to carry out the attack, they may not be motivated to run the risk of trying to instigate it.
This is, of course, what the introduction of awareness and strategy brings out. But, as I argued earlier, relying on these epistemic criteria – which demand that potential team members need knowledge of each other’s actual mental states – constrains the relevant domain of possible worlds too narrowly. My claim is that neither awareness nor strategy are individually necessary for holding power as a social group. Instead, the existence of social structures that bring about unequal distributions in status and credibility can itself be sufficient for members of some groups to hold dominating power over others. To see how and why this is the case, we need to see how non-domination can account for the informal basis of social institutions. This is what I do in the next section.
Social norms, epistemic injustice and responsive control
In the literature on non-domination, there are frequent references to the institutions of slavery, apartheid and coverture. What these have in common, is that they refer to social institutions underpinned by law. This makes non-domination capable of grounding a freedom-based critique of social and legal orders that do not offer equal citizenship to everyone subjected to their laws. However, most proponents of non-domination want their theory to have wider application, so that it can diagnose domination between people who enjoy status as equal citizens. Racial and patriarchal oppression persists in virtually all societies despite official recognition of formal equality before the law. These forms of oppression manifest in the ability of some to interfere in the acting of others, granted not by unequal law but instead by widespread social norms. However, some object that non-domination is incapable of diagnosing domination under conditions of equal citizenship. Given the claim that only law can guard against domination, the objection goes, social norms are too weak to play any role in securing relations of non-domination. As I will now argue, this kind of argument fails once we adopt a systemic conception of domination. This allows us to construct a response to the coalition problem that retains non-domination’s critical potential, which counts as a reason for favouring the systemic conception.
Unsurprisingly, the argument that non-domination is bound to treat social norms as irrelevant for freedom appears in both Dowding and Simpson. According to Dowding, non-domination cannot distinguish in terms of freedom between two societies that (…) had the same constitutions, regulatory institutions, and enforcement practices (…) even if in the first one class of people (say women) had continually to resort to the law if they wanted their liberties respected, whereas in the other they rarely had to. After all, in both societies, the men have the same capacity to act, they have the same physical powers, and the legal constraints are identical. The only difference in the two societies is the dispositional nature and habitual behaviour of men. (2011: 312)
It seems implausible that women in a society where laws in favour of gender justice are widely disregarded are no less free than those who live in a society where the same laws are also supported by anti-patriarchal social norms (cf. Krause, 2013). So does the claim that the bystanders’ relation to the Dalit is unchanged by the fact that ‘caste norms have no social power’. So why should we think that the non-domination conception is bound to deliver these results? Here is one suggestion. A person’s ability to do what they are physically capable of depends on their motivation. Individuals can be motivated to do most things, given the right incentive. (Therefore) freedom from domination requires external constraints on action in the form of the rule of law: the existence of coercive state powers works on individual wills, posing a motivational block on what individuals may be physically capable of doing. Social norms, by contrast, are not enforced by coercion. They are enforced by mechanisms of social disapproval. These are psychologically powerful but will ultimately fail to render individuals unresponsive to incentives. Further, unlike laws, norms have their source in dispositional and habitual behaviour. Thus, they do not seem to be sufficiently external. Although individuals may unthinkingly be reproducing social norms, they have a choice about whether or not to do so.
The problem with this account is that it underplays the relation between the social norms, on the one hand, and the rule of law, on the other. Social institutions can be supported by norms and laws, and although these have different enforcement mechanisms, they both function to bring about the results and relations that make up the relevant institution. A social institution with significantly less support from either source will bring about those results and relations less stably. Against the critics, this does have consequences for freedom from domination. To see this, consider first what kind of protection the rule of law grants to citizens who enjoy the same legal status. Andreas Schmidt (2018) has offered a useful taxonomy of the various ways in which agents can hold the control required for posing an external constraint of the wills of others. An agent holds preventive control if they can prevent others from interfering; holds abortive control if they can remove a constraint on their action after it is already in place; and holds responsive control if they have means to respond to an interference after it has happened (Schmidt, 2018: 180). The rule of law ensures that the state has a monopoly on the use and regulation of coercive force, which means that preventive and abortive control falls within the domain of its competence. 15 The main resource that the law grants individual citizens, then, is responsive control. If and when someone suffers illegal interferences with their acting, they can call on the state to have the interferee punished. This responsive control can survive the individual – cases of murder are still investigated on behalf of the deceased.
To see how this complicates the claim that norms cannot impact an individual’s domination, consider now how we should understand the Dalit’s situation in the second variation of that example. According to Simpson, the Dalit is dominated and the main source of his unfreedom is the potential team composed of the angry Brahmin and the bystanders. We know from the description of the case that the involved parties are in a remote place and can therefore assume that no representatives of the state are present to exercise preventive or abortive control over what, we must also assume, is an illegal attack. We should also assume that the consequences for those who engaged in the attack would be serious, were they to be found guilty in a court of law. 16 Further, since caste norms hold no social power, there exists no reason why individuals would single out the Dalit as a legitimate subject for an attack. Acts in favour of the subordination of Dalits are not reproduced in action. This has significant consequences for the plausibility of the claim that the Dalit is, indeed, dominated on the description of the case.
Being bound by a social norm makes our preferences conditional on the acting of others. But, as Cristina Bicchieri (2006: 30–31) explains, that preference is not simply conditional on how we believe others will react to our acting; it is also a function of how we believe others will act in relation to the content of the norm. Thus, when people are under a norm, they have normative expectations that others will sanction norm transgressions and empirical expectations that others will follow the norm themselves. These empirical expectations provide a crucial link between social norms and laws in circumstances – like the second variation of the Dalit case – where there exists a general expectation that people uphold the law and that all individuals enjoy equal social status. This link provides an explanation for why the convergence of individual motivation is insufficient for the feasibility of collective action that runs counter to the law.
When norms are directly tied to the law, there are material as well as ‘immaterial’ consequences to norm transgression. Given that individuals live under the rule of law, in conditions where each party is seen as an equal before the law (caste norms have no social power), part of obeying the norm of following the law is to support the law’s application. Each party will expect that the others will not only refrain from joining the attack, but that they will be of assistance in the process of the state exercising responsive control on behalf of the Dalit. In particular, each will expect that the others will offer testimony to the state. This means that, even when each party is in fact motivated to act, they will be in what Bicchieri calls a state of pluralistic ignorance (2017: 42–43). Such a state obtains when (i) individuals engage in social comparison with their peers that is based on inferences from their acting, since the ‘true distribution of their beliefs and preferences’ is hidden; (ii) others’ behaviour (or the results of that behaviour) are observable; (iii) no transparent communication is possible; and (iv) we assume that, unlike us, others act in accordance with their beliefs. Bicchieri’s analysis is directly relevant to the situation around the well. Conditions (i) and (iv) are general facts about social life. 17 Condition (ii) obtains because people are not reproducing the norms underpinning the caste system, which would justify the attack. Lastly, condition (iii) obtains because, if any of the bystanders were to voice their true preference, they would face legal sanction: it is illegal to instigate attacks and all bystanders are under norms to assist the state in exercising responsive control. This shows why it is implausible to claim, as Simpson does, that the Dalit it dominated in the second variation of that case. In conjunction with the laws that are in place to secure the social institution of equal citizenship, norms can act as an external constrain on individual wills by reinforcing the rule of law. The claim that norms are insufficient for freedom from domination does not entail that norms are not necessary.
To bolster this conclusion, see how the present account provides us with the resources for arguing that the Dalit would be dominated in the absence of the norms in favour of supporting the rule of law. 18 I have shown why people would be unwilling to engage in random attacks if they were confident that their actions would be punished. What is lacking, then, from a convincing explanation of the Dalit’s domination is an account of what could make it the case that the bystanders would not be punished for their attack. But, as I will now show, such an account necessarily entails that the Dalit was already dominated. The rule of law can only secure freedom from domination in the absence of systemic conditions that distort the law’s equal application. When such conditions obtain, members of systematically disadvantaged groups lack the effective ability to draw on the state to exercise responsive control. They suffer systemic domination.
Diagnosing lingering systemic conditions of domination after the ‘official’ rejection of dominating practices is a key concern of the epistemic injustice literature. The innovation of Miranda Fricker (2007) – who first coined the term – was to provide a language for theorising how transmissions of knowledge happen according to patterns that systematically disadvantages certain groups. When individuals seek out information about an event, they need to rely upon the testimony of those who were present and, in so doing, they engage in implicit credibility judgements of the speakers. In this process, ‘testimonial’ epistemic injustice happens when a speaker is given less credibility than what she is owed, due to prejudice about her identity group on the part of the hearer (Fricker, 2007: 18). In this case, her testimony’s value is deflated solely on the basis of her (perceived) group membership.
The concept of epistemic injustice has crucial application for analyses of domination (cf. Bohman, 2012; Fricker, 2013). Take the Dalit’s situation. 19 Since the well is in a remote place, the Dalit’s protection under the law directly depends on his ability to exercise responsive control. But his ability to do this depends on his epistemic standing. His testimony must be believed, and it must be believed over countervailing testimony offered by the Brahmin and the bystanders. The presence of epistemic injustice thus revokes the Dalit’s ability to safely rely on the state’s protection. Was he just drinking from the well? Did he not do anything to provoke the Brahmin? Why would a respectable group of people randomly attack someone? These questions, indicative of an inferior positions in the economy of credibility, are crucial in the processes that ensure the continued domination of certain groups. The presence of epistemic injustice removes the third condition of what would otherwise be a state of pluralistic ignorance by reducing the cost of transparent communication. This provides a plausible explanation of why the Brahmin in Simpson’s example has already decided to ‘test the water’ (cf. Lovett and Pettit, 2019: 377). He knows that, even in the case where he is unsuccessful in his attempt to enlist others, the status imbalance between him and the Dalit will distort the latter’s ability to exercise responsive control. 20
The presence of social norms that ground epistemic injustice prohibit the ability of systematically disadvantaged groups from exercising responsive control. 21 This makes them vulnerable to illegal interference and, on the view defended here, socially unfree. This remains the case even when those who are reproducing the norms are unaware that they are doing so. What matters is the existence of social norms that render some unable to safely rely on the law to exercise responsive control, since this makes them vulnerable to those who can exploit their advantage before the law. This analysis shows how domination by potential groups can persist in cases where neither the awareness nor strategy conditions are fulfilled. The function of the social norms grounding epistemic injustice is to remove the cost of acting to satisfy awareness and strategy. It is these norms, and not the individual knowledge of the others’ mental states that they can bring about, that are relevant for assessments of domination: they provide the circumstances against which assessments of feasible individual and collective action are to be assessed. However, as Fricker’s exposition of epistemic injustice makes clear, this does not make the diagnosis of domination purely functional. 22 The norms in question affect social freedom because they distribute social power in ways that, in turn, affect individual encounters. In a racist society, a member of a racialised minority group is vulnerable to members of the majority group because the latter could get away with an illegal interference in their acting. This, on my view, is a fact that bears directly on the freedom of that individual.
The domain of possible worlds that is relevant for assessments of social freedom ought, therefore, to be grounded in contextual analyses of operative social norms. Doing this dissolves the coalition problem because it explains how pluralistic ignorance renders the convergence of individual wills insufficient for feasible collective action. Further, it does this in a way that allows us to diagnose domination more broadly than does Lovett and Pettit’s alternative. A slave with three masters is dominated because the legal and social norms of that institution supply the room for voicing desires to interfere and for experimenting about how to carry out their interference in the absence of a clear path. This remains the case even if the slave were somehow oblivious of his circumstances and, as a result, would not fail the eyeball test.
Conclusion
An account of freedom as non-domination that is properly attuned to the systemic aspects of power can dissolve the coalition problem in a way that coheres with our experience of who hold power, and why. This provides strong reason to favour it over accounts that conceptualise domination in a primarily agent-relative way. It is the systemic features of society that dictate who can feasibly combine their individual resources in a way that impact the freedom of others. These systemic features of society can, further, be studied in ways that does not require us to moralise the concept of domination: social norms are facts about social life and highlighting their consequences on abilities to act does not require making substantive value judgements. Such norms exist at local, domestic and global levels, which means that the study of systemic domination has a large scope of application. The theoretical advantages of turning to a systemic analysis are not local to the coalition problem. Dorothea Gädeke (2020) has recently demonstrated that systemic conceptions of domination hold the potential for solving other pervasive philosophical problems for freedom as non-domination, including the charge that it is inevitably moralised. The argument in this article should be read as strengthening the case made by Gädeke – and Laborde (2008, 2010) and James Bohman (2012) before her – in favour of freedom as non-domination’s unique critical potential when it is attuned to the systemic conditions under which individual and collective agents act.
One might worry, however, about the consequence of this move for the analysis of social freedom as a whole. Is it not problematic to claim, as I do, that when we engage in analyses of the freedom of a person, we should be centrally concerned with macro-level accounts of the social structures of which they are part? Does this not commit me to a suspicious methodology that treats aggregate phenomena as explanatorily prior to the individual actions of which they are composed? These questions are obviously too large to satisfactorily answer here but let me just note the following in response. If a theory is concerned with making explanatory, as opposed to ontological, claims then it is unproblematic to accept that all social facts supervene on individuals and their actions while at the same time insisting on treating social structures as such facts (List and Spiekermann, 2013). The question, then, is what we want our theory of social freedom to do. It is not clear to me why we should want a theory of social, as opposed to metaphysical, freedom to be geared towards making ontological claims. Indeed, one way to rephrase the argument made in this article is the following: simply pointing to what people could ‘ontologically’ do together is a misguided way of explaining who is free to act in the social world, and why.
Footnotes
Acknowledgements
Previous versions of this article were presented at the Nuffield Political Theory Workshop, and at the Practical Philosophy Seminar at the University of Oslo. I want to thank the participants at these events and Tom Simpson for very helpful discussion of the ideas and arguments in the article. I also want to thank Robert Huseby, Henrik Kugelberg, Cécile Laborde, Reidar Maliks, Hans Robin Solberg, Johan Trovik, Hannah Voegele, and two anonymous reviewers for this journal for written comments that much improved the article. I owe special thanks to Cécile, who commented on several drafts of the article and whose advice throughout the writing process has been invaluable.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was generously supported by an Aker Scholarship.