2.a The proposal

A simple proposal says a collective can legitimately make an overdemandingness complaint just in case the enactment of its purported obligation would be overdemanding for at least one member. Call this proposal Membership Overdemandingness.Footnote ⁸ This is a tempting view of the introduction’s examples: perhaps curators could complain if the Louvre devoted its proceeds to poverty relief; perhaps the owners could complain if Mom-and-Pop Gas devoted itself to electric cars; perhaps citizens could complain if Sweden extended the franchise to foreigners. If curators, owners, and citizens are “members” of art galleries, gas stations, and democratic states (which seems plausible), then perhaps the overdemandingness complaints mentioned in the introduction rise and fall with the overdemandingness complaints of members.

Our guiding question was: Can collective agents legitimately make overdemandingness complaints that aren’t equivalent to (a bundle of) individual overdemandingness complaints? Membership Overdemandingness might seem to allow an equivalence thereby evading the question. But a biconditional doesn’t entail equivalence. Perhaps a collective’s overdemandingness complaint arises if and only if member(s) have overdemandingness complaints while being distinct from them. Such distinctness could arise via two routes: the collective’s complaint might be multiply realisable in members’ complaints, and/or there might be systemic relations between members’ complaints (Elder-Vass Reference Elder-Vass2007; List and Spiekermann Reference List and Spiekermann2013). It’s worth exploring these, to see how Membership Overdemandingness might work.

Unfortunately, the multiple realisability route won’t get us distinctive collective-level complaints. If the multiple realisability of some collective-level phenomenon is to render it distinct from its individual-level realisers, then the different potential individual-level realisers must be of different types, not just different tokens of the same type. Consider the analogous case in philosophy of mind: suppose my pain is always realised by C-fibres firing in my brain, though different C-fibres fire on different pain-occasions (because C-fibres degenerate and are replaced by new ones). Then, pain is equivalent to C-fibres firing. For pain to be nonequivalent to (i.e., distinct from) its neural realisers, there must be different types of potential neural realisers. According to Membership Overdemandingness, collective overdemandingness complaints are always realised by the same type of thing: individuals’ overdemandingness complaints. This makes collectives’ overdemandingness complaints equivalent to individuals’ complaints. So this isn’t how to understand Membership Overdemandingness.

The second route to nonequivalence is on firmer ground. This route says collectives have distinct complaints because there are systemic relations between members’ complaints, even though the collective’s complaint rises and falls with members’ complaints.Footnote ⁹ The systemic relations produce a network of individuals’ complaints, where the network is over and above the bundle of complaints in the network. The network is “over and above” because the relations between the complaints are also part of the network. Thus we can identify a collective’s complaint with a systemic network of member complaints.

For example, perhaps Sweden’s complaint against franchise-extension arises if and only if there are citizens’ complaints against franchise-extension such that the citizens’ complaints are related by having the same source. The source is the laws and norms of extended-franchise Sweden. This common source systemically relates the members’ complaints to each other, rendering members’ complaints components in a holistic network of complaints that constitutes Sweden’s complaint. The network of complaints is more than the sum of its member-complaint parts. Thus Membership Overdemandingness gives rise to distinctive collective-level complaints via a tweaking of the biconditional: a collective can legitimately make an overdemandingness complaint just in case the enactment of its purported obligation would be overdemanding for more than one member in systemically related ways.

However, the proposed biconditional doesn’t hold. A collective’s obligation being overdemanding on more than one member (in systemically related ways) is neither sufficient nor necessary for that obligation being overdemanding on the collective. Even if a collective’s obligation is overdemanding on every member (in systemically related ways), that’s neither necessary nor sufficient for the obligation to be overdemanding on the collective. I’ll now show this.

2.b Insufficiency

First, a collective’s obligation being overdemanding on members (even with systemic relations between their complaints) is not sufficient for its being overdemanding on the collective. This is clearest in organisations with altruistic goals, such as nonprofit organisations pressing for justice. The collective’s ardent pursuit of justice might be extremely demanding on members because members have goals other than justice (e.g., leisure). If the collective’s only goal is justice, then any action members can take toward that goal will not be overdemanding for the collective—considered as an entity picked out by its distinct bundle of (functionally characterised) beliefs, preferences, hopes, regrets, and intentions, and the procedure that produces that bundle when enacted by members. Justice-pursuing actions that infringe on members’ private lives are a boon to the collective: these actions pursue the collective’s only goal. Yet such actions might be overdemanding for members—perhaps even for every member—with complaints being systemically related via a common cause (namely, the organisation’s justice goal).

Perhaps the organisation’s obligation does dissolve if that obligation is overdemanding on one, some, or every member. But that dissolution should not be conceptualised as the obligation’s being overdemanding on the organisation. Instead, if the organisation’s obligation dissolves due to its overdemanding effects on members’ private lives, then the obligation dissolves due to detrimental effects on third parties. Given that (ex hypothesi) members’ leisure is not a goal of the organisation, those detrimental effects are not effects on individuals qua members. They are, instead, effects on individuals qua third parties, considered external to the organisation. If an obligation of any agent has overly detrimental effects on external parties, then the obligation may dissolve. But it’s false that the external agents’ complaints make the obligation overdemanding. For example, if my keeping a promise to my friend would cause you great suffering, then your suffering may dissolve my obligation. But your suffering isn’t sufficient for my obligation being overdemanding on me. Things are analogous for members’ private suffering and the dissolution of organisations’ obligations.

One might say: if the organisation valued justice, then it would recognise that member burnout is detrimental to justice. It would therefore value members’ well-being. Thus, infringements on members’ well-being would be internal costs to the organisation (not external costs on third parties), and high infringements on members’ well-being would indeed be sufficient for an organisation’s overdemandingness complaint. Membership Overdemandingness would be vindicated.

In reply: for this to create a legitimate group-level overdemandingness complaint, we would need an argument for upholding the organisation’s justice goal. After all, according to the objection, the organisation values members’ well-being only as a means to justice. If the organisation values members’ well-being intrinsically, then we’d need an argument for upholding that goal, too. I’ll consider such arguments in the next section, under the heading of upholding organisations’ permissible deep preferences.

A different defence of Membership Overdemandingness runs as follows. Perhaps members’ overdemandingness (e.g., lack of leisure) is a cost internal to the collective, and therefore sufficient for a collective overdemandingness complaint, because members are not really external agents to their collectives. When a collective has an obligation, this plausibly entails obligations for at least one member (Collins Reference Collins2019, chap. 7). Call these membership obligations. The costs of membership obligations (i) must be imposed upon members by themselves (which is a feature of overdemanding obligations [McElwee Reference McElwee2017]) and (ii) are not held by members qua third parties (so the obligations’ demands are internal to the collective). The sufficiency half of Membership Overdemandingness appears vindicated.Footnote ¹⁰

To respond, we must distinguish the different parts of an individual’s life: parts internal to, and external of, her collective. When an individual discharges her overdemanding membership duty, she inhabits one part, and imposes costs upon herself in another part. Although the cost imposer is internal to the collective, the cost experiencer (the part that wants leisure) is external. Here I follow Carol Rovane’s ontology, under which “when human beings achieve rationality together at the level of the whole group, this tends to produce a certain kind of rational fragmentation in their lives, because not all of the thoughts and actions associated with their brains and bodies proceed from the same point of view—some proceed from the group agent’s point of view, while others proceed from a point of view that is somewhat smaller than human size” (Reference Rovane2014, 1665; similarly Rovane Reference Rovane1998; Rovane Reference Rovane2004).

Once we separate the cost imposer from the cost experiencer—though both are housed within one human life—we see that the cost experiencer remains external to the collective, even though the costs are self-imposed from the perspective of the whole human life. But the costs are not self-imposed by the collective upon the collective, or by the individual-qua-member upon the individual-qua-member. The complaints of the individual-qua-private-person are insufficient for complaints of the collective because this particular part of the individual’s agency is external to the collective.

Thus, a collective obligation’s being overdemanding on “a member” is triply ambiguous: Is it overdemanding on member-qua-member, member-qua-private-person, or member-qua-whole-human? The member-qua-member is the only one whose complaints are truly internal to the collective. But the member-qua-member pursues the collective’s goals. So, any complaints made from within this role must be complaints that the collective’s goals are infringed, not complaints about leisure. But section 3 will argue that infringements of a collective’s goals don’t produce legitimate overdemandingness complaints. Meanwhile, complaints of members-qua-private-persons or members-qua-whole-humans are external to the collective. Member overdemandingness is insufficient for collective overdemandingness.Footnote ¹¹

2.c Nonnecessity

Member overdemandingness is also unnecessary for collective overdemandingness. Complaints of members-qua-private-persons and members-qua-whole-humans can contradict the goals of a complaining collective, as the justice example demonstrates. So it would be strange indeed if these external parties’ complaints were necessary for a collective’s complaints. Meanwhile, complaints of members-qua-members are complaints made with the collective’s goals in view. Such member complaints might be necessary for collectives’ complaints, but assessing the legitimacy of those complaints requires us to assess whether the collective’s goals should be protected. That’s the task of the next section.

This might look suspicious. Above I said collectives’ obligations entail members’ obligations and that member actions are incorporated by collectives. Plausibly, when members act within, because of, and while performing their role, their actions are incorporated by the collective (Collins Reference Collins2018). If that’s right, where could collectives’ overdemandingness complaints possibly come from if not from the member actions that constitute collective actions?

From the frustration of the collective’s preferences. Suppose all Sweden’s members become indifferent to whether Sweden extends its franchise abroad, including all costs this would impose on them. Nonetheless, suppose (arguendo) that such franchise extension would undermine Sweden’s sense of community. Then extending the franchise would undermine one of Sweden’s preferences, namely, retaining its sense of community. It would be overdemanding on Sweden itself, even if members individually endorse the action (and the resulting loss of community). This example provides a concrete demonstration that member overdemandingness is unnecessary for collective overdemandingness. To attempt to vindicate collectives’ overdemandingness complaints, we must turn to the collective’s preferences.

3.a The proposal

The preceding section emphasised that collectives have distinct preferences, whose content can depart from all members’ preferences. Given this, perhaps an obligation places a demand on a collective just in case it frustrates one of the collective’s preferences. And an obligation is overdemanding on a collective just in case the frustrated collective preference is a “permissible deep” preference. Call this proposal Deep Preferences.

A deep preference is long held by its bearer, grounds many of the bearer’s other preferences, and is held by the bearer as amongst the bearer’s most important preferences. This is a high bar for a preference to meet, including objective and subjective elements. The height of the bar is intentional: the more conditions a preference must meet to count as deep, the more compelling is the idea that it’s overdemanding to frustrate such preferences. We should understand permissible as permissible when considering aspects of morality other than overdemandingness. This prevents us from begging any questions about which preferences are rendered permissible by overdemandingness complaints, while capturing the thought that morally abhorrent preferences (such as racist preferences) cannot serve as legitimate complaints against obligations that would frustrate those preferences.Footnote ¹²

Deep Preferences has three attractions. First, it respects the previous section’s result: member overdemandingness is neither necessary nor sufficient for collective overdemandingness. A collective can hold a permissible deep preference that no member holds, and every member can hold a permissible deep preference that the collective doesn’t hold. We saw this with the art gallery that prefers to display contemporary art while all members prefer the gallery to display noncontemporary art.

Second, deep preferences seem present in the introduction’s examples: an art gallery’s deep preference is to display art; a gas station’s deep preference is to sell gas; a state’s deep preference is to retain a sense of community. Perhaps these preferences aren’t permissible when considering aspects of morality other than demandingness. Arguably, the gallery’s preference distracts from more morally important goals, the gas station’s preference is positively harmful (even if it also benefits motorists), and the state’s preference disrespects foreigners affected by its laws. But—the advocate of Deep Preferences might respond—that simply shows that these specific examples aren’t legitimate complaints. Still, a collective’s permissible deep preferences should be protected. In fact, Deep Preferences provides a good error theory of the introduction’s examples: we are tempted to grant these overdemandingness complaints because collectives’ deep preferences are at issue; we err in assuming the deep preferences are permissible when considering aspects of morality other than demandingness. So Deep Preferences does well at addressing these examples.

A third attraction is that deep preferences seem operative in the overdemandingness complaints of both individual and collective agents. This attractively unifies our theory of overdemandingness: our theory can apply to collectives just as it applies to individuals. An individual’s deep preferences might include “being an excellent teacher and researcher, being a dedicated parent, maintaining lifelong friendships, or seeking new cultural experiences,” while a collective’s deep preferences might include “searching for oil” or “worshipping [and] proselytising.”Footnote ¹³ Perhaps all these preferences block moral obligations via overdemandingness (Collins and Lawford-Smith Reference Collins and Lawford-Smith2016, 52–53).

The third attraction of Deep Preferences is its Achilles’ heel. By raising the prospect of a unified overdemandingness theory across agent types, it invites us to apply to collectives our justifications for protecting individuals’ permissible deep preferences. If the justifications don’t apply, this is bad news for collective overdemandingness complaints. To assess this, I will consider whether three prominent justifications for protecting individuals’ permissible deep preferences apply to collectives. These proposals come from Garrett Cullity, Liam Murphy, and Samuel Scheffler and Bernard Williams. For argument’s sake, I assume these justifications satisfactorily protect individuals’ deep preferences.Footnote ¹⁴ Still, they don’t extend to collectives. Perhaps there is a sui generis justification for protecting collectives’ deep preferences, but my discussion will leave that looking far-fetched.

2.b Reciprocal obligations

Garrett Cullity (Reference Cullity2004) argues morality cannot infringe individuals’ permissible deep preferences because we judge that individuals are sometimes obliged to help other individuals pursue similar preferences. For example, if I can help you contact a long-lost friend at little cost to myself, then I have a duty to do so. This intuition, Cullity argues, is on par with intuitions used to defend demanding moral conclusions, such as Singer’s (Reference Singer1972) intuition that one must sacrifice one’s shoes to save a drowning toddler, from which Singer argues that one must sacrifice a lot to ending poverty. According to Cullity, my obligation to help you contact your friend implies you are permitted to contact your friend. But if you’re permitted to contact your friend, consistency requires that I’m permitted to pursue similar projects. In general, individuals are permitted to pursue whichever projects others would have obligations to help them pursue. This includes at least some permissible deep preferences.

The problem is this. We don’t tend to judge that collectives are obliged to help one another pursue permissible deep preferences, or that individuals are obliged to help collectives pursue permissible deep preferences. That is, Cullity’s envisaged reciprocal obligations of project-assistance (with concomitant permissions of project-pursuit) don’t include collective project-pursuers or collective project-assisters.

Of course, some collectives have reciprocal legal obligations. For example, sometimes tech companies have obligations to share code and large firms have obligations not to compete with small businesses.Footnote ¹⁵ But these are not reciprocal obligations to help one another pursue projects. Instead, these obligations are grounded in social utility considerations: by requiring code sharing or restricting competition, we bolster gross domestic product, or human happiness, or so on. And our intuitions about these obligations do not have the same “simply intuitive” status as Cullity’s friend-helping or Singer’s toddler-saving. Cullity’s argument relies on that status.

Likewise, sometimes collectives are obliged to help individuals pursue projects. Perhaps my employer is obliged to provide me parental leave to help me pursue my project of raising children. But, again, this can’t grant my employer permissions via Cullity’s strategy. The strategy applies only when the projects of the assister and assistee are similar. It’s about reciprocity: I’m permitted to pursue projects like those I’m obliged to help others pursue. But the kinds of projects collectives are obliged to help individuals pursue (parenthood, leisure, work–life balance) aren’t the kinds of projects collectives can pursue because collectives aren’t biological creatures. Reciprocal obligations don’t protect collectives’ permissible deep preferences.

2.c Fairness

Liam Murphy (Reference Murphy2000) provides another route to protecting individuals’ permissible deep preferences: fairness.Footnote ¹⁶ In brief: if everyone was willing to do what morality required, then a fair division of labour wouldn’t require individuals to sacrifice deep preferences. The deep preferences of compliers are infringed only because noncompliers refuse to do their fair share. Murphy’s strategy allows us to protect compliers’ permissible deep preferences if, and because, the demand to forgo them would treat compliers unfairly and disrespectfully relative to noncompliers.

Do collectives deserve fairness and equal respect? In the literature on collective intentionality, agency, and responsibility, almost all philosophers are “normative individualists”: although collectives might have intentionality, agency, and responsibility, only individuals are of intrinsic moral worth.Footnote ¹⁷ Even those who endorse rights for collectives view those rights as a necessary means to some end, such as social utility or collectives’ accountability (Smith Reference Smith2018). And even those who argue collectives deserve moral consideration do not require that we treat them fairly, or with equal respect, relative to individuals (Silver Reference Silver2019).

That said, even without a requirement to treat collectives fairly and with equal respect relative to individuals, there might be a requirement to do so relative to other collectives. This is compelling when two collectives are (in some sense) the same type. For example, it might be problematically unfair, or unequally respectful, if funding bodies applied higher teaching or research standards to your university than my university. And—unlike collectives’ obligations to one another under Cullity’s proposal—this needn’t be understood as a means to gross domestic product or human happiness. Perhaps it’s intrinsically wrong to treat universities differentially in these ways.Footnote ¹⁸ Could the need for fairness and equal respect protect collectives’ permissible deep preferences, shoring up their overdemandingness complaints?

I suggest not. Under Murphy’s proposal, individuals’ permissible deep preferences are protected conditional on the contingent, partly empirical fact that fair shares don’t infringe individuals’ permissible deep preferences. This is less plausible for collectives. Consider oil companies: treating one oil company fairly (and equally respectfully) relative to others plausibly means demanding that all oil companies redirect their operations. Likewise for states extending the franchise. The same even goes for art galleries: if it’s fairness within each collective-type that matters, then we need some independent story about why fairness amongst art galleries should be expected to preserve the display of art. Fairness and equal respect might spell the (fair and equal) end of all art galleries. Fairness and equal respect within each collective-type does not imply nondemanding obligations.

But perhaps morality demands fairness and equal respect across all collective-types. If so, some collectives’ deep preferences might be protected. But, I suggest, morality does not demand fairness and equal respect across all collective-types: we typically think, for example, that states are more sacrosanct than corporations or art galleries. If there is a sense of fairness that demands fairness across all collectives, it’s a sense of fairness that permits highly differentiated (and, sometimes, highly demanding) obligations across different collective-types. The upshot: to protect collectives’ permissible deep preferences, we need something more robust than fairness or equal respect for collectives.

3.d Projects and points of view

A third justification provides such robustness. It says deep preferences are intrinsically valuable from the perspective of ground projects or the personal point of view. According to Bernard Williams (Reference Williams and Williams1981, 12–13), ground projects are “closely related to [one’s] existence,” “give meaning to … life,” and provide “a reason for living.” According to Samuel Scheffler (Reference Scheffler1982, 56, 77), the personal point of view is the perspective from which “projects are undertaken,” “plans are developed,” “events are observed,” “life is lived,” and “a locus relative to which harms and benefits can be assessed.” Perhaps most importantly, the personal point of view is “a source for the generation and pursuit of personal commitments and concerns that is independent of the impersonal perspective” (1982, 57).Footnote ¹⁹

For Williams and for Scheffler, legitimate overdemandingness complaints derive from the intrinsic value of ground projects or the personal point of view, both of which are partly constituted by deep preferences. For Scheffler, this value permits individuals to give more weight to their personal point of view than the impersonal point of view. For Williams, ground projects seem to provide a hard constraint against moral theories that require individuals to make decisions purely from the impersonal point of view. Either way, the weight or constraint are not conditional on contingent empirical facts about what needs to be done to make the world better, as under Murphy’s view. If collectives have ground projects and personal points of view, then perhaps Scheffler’s and Williams’s stories apply to them.

Before concluding this, we need to ask why ground projects or personal points of view matter. Several philosophers have argued they matter only insofar as their bearers are phenomenally conscious, which rules out collectives (Hess Reference Hess2013; Pasternak Reference Pasternak2017; Berkey Reference Berkey2019, 126). Yet Williams and Scheffler do not evoke consciousness. They do not emphasis that humans have negative qualia when their ground projects or personal points of view are stymied; they do not emphasise “what it is like” to have ground projects and points of view; they do not discuss feelings. Williams even distances project pursuit from feelings: ground projects are a cornerstone of his rejection of utilitarianism, which (in Williams’s view) focuses too much on producing good feelings (Reference Williams1963, 112–13; Reference Williams and Williams1981, 8). Instead, ground projects and personal points of view are connected to agency: as reflected in the quotes above, ground projects and personal points of view provide reasons, meaning, projects, plans, assessments, commitments, and concerns.

Thus, we need not rely on consciousness to resist applying Williams and Scheffler to collectives. There’s a different reason for resistance: collectives’ ground projects and personal points of view are not produced in the right way. They are produced via a procedure that is thrust upon the collective from the outside when the collective is established. To be sure, the collective’s procedure is maintained internally via members-qua-members’ commitments, inputs, and actions. But internal maintenance can happen only once the procedure is established. The establishment of a collective’s agency—and the establishment of the procedure by which its agency is exercised—is a process enacted upon the collective by individuals (or other collectives) who, at that establishment moment, should be conceived of as nonmembers (though perhaps soon-to-be members). A collective’s decision-making procedure is established by agents other than the collective itself.

Contrast individuals. When individuals are created, they are not instantly agents. Individuals’ agency develops through a complex interplay between the individual and her environment. The creators (parents) lack the power to intentionally impose particular procedures, so parents cannot shape individuals’ projects or points of view in the way collectives’ creators can. An individual may lack decision-making procedures altogether—certainly, her projects and point of view are not driven by decision-making procedures designed by others and imposed upon her from the outside when she is created.Footnote ²⁰ This is reflected in Williams’s emphasise on convictions (1963, 116). Humans’ convictions are arrived at by no procedure. Contrastively, collectives have convictions only via a procedure that was established, and is maintained, by others.

So individuals have fuller agency over their ground projects and points of view than do collectives. Full agency seems to be at the core of Scheffler’s and Williams’s proposals. If fullness of agency is what matters, then we have weaker reason to value collectives’ ground projects and personal points of view (and the deep preferences that partly constitute those projects and points of view) than we have to value individuals’.

At this point, one might wonder whether collectives’ agency is so attenuated that they cannot bear moral obligations at all.Footnote ²¹ If so, the answer to my title question is “nothing”—not, as I’ve said, “an awful lot.” To respond, we must consider the function of moral obligations. From the first-personal perspective (when one has an obligation), obligations paradigmatically function as fitting inputs into one’s decision-making about what to do or be. From the second-personal perspective (when one is owed an obligation from another), obligations paradigmatically function as demands or claims. From a third-personal perspective (when another owes an obligation to someone else), obligations paradigmatically function as advice or guidance to the obligation bearer. All three functions presuppose merely that the obligation bearer has the ability to make decisions that are sensitive to obligations. Footnote ²² Collectives have that ability. So they are agents in a full enough sense to bear obligations—even if the ability was imbued in them by others such that they lack the complete fullness of agency at issue in the Williams–Scheffler view.

Could there be a fourth justification for protecting collectives’ deep preferences? The prospects are grim. We’ve seen that almost all philosophers deny collective consciousness (or its value), that collectives’ agency is made and maintained in a less than fully autonomous way, and that fairness and equal respect are uphill routes to nondemanding collective obligations. We’ve seen that our intuitions about collectives’ role in our moral world don’t endorse preference protection. I conclude that collectives’ permissible deep preferences are up for moral grabs.Footnote ²³

4.a The proposal

Yet collectives’ moral obligations are sometimes limited. These limitations provide a reinterpretation of our intuitions about art galleries, gas stations, and states.

To begin, notice that collectives have agency over only the part of the world regarding which their decision-making procedures can process inputs and make decisions. That part of the world may not include everything regarding which there are moral principles; a collective’s decision-making procedure may thus be constrained. It might be constrained in two ways: by its form (e.g., voting, committees, discretion, bargains, norms, etc.) and by its fundamental ends (i.e., the ends the collective cannot but pursue, at least within a particular time period). I’ll use “constitutional constraints” to refer to the aspects of a decision-making procedure’s form and fundamental ends that impose limits on the collective’s domain of agency within some time period. (This point about the narrow domain of [some] collectives’ agency is separate from section 3’s point about the origins of [all] collectives’ agency.)

My suggestion is that constitutional constraints place limits on collectives’ abilities. We therefore violate “obligation implies ability” when we ask a collective to do what it is constitutionally constrained from doing. Collectives’ legitimate overdemandingness complaints transmute into complaints that the collective is unable to perform the obligation. But inability is distinct from overdemandingness. The result is that overdemandingness per se is never an excuse for a collective not to perform an obligation. The closest we get are complaints that obligations violate constitutional constraints. These are best understood as complaints that “obligation” implies “ability” and the collective lacks the relevant ability.Footnote ²⁴ Call my suggestion Constitutional Constraints.

We can contrast constitutional constraints with deep preferences. Recall: a deep preference is long held, undergirds other preferences, and is viewed by the collective as important. But many deep preferences are overridable in the collective’s practical reasoning. A collective’s constitution, by contrast, is the framework within which preferences are formed, networked together, and selected as important. The constitution also contains some preferences, but these are stronger than deep preferences: the preferences in the constitution target fundamental ends, which are ends the collective cannot but pursue, at least within a particular timeframe.Footnote ²⁵

For example, suppose some friends establish a basketball team for the local competition. This happens by one friend asking the others if they want to comprise a team, the others saying yes, and one doing the registration. This establishes the collective and, with it, the decision-making procedure’s form and fundamental end. The procedure’s form is conversation-based consensus. The procedure’s fundamental end is playing basketball. This is the end the collective cannot but pursue, at least in the short term. This example demonstrates that a constitution doesn’t necessarily have legal status, such as articles of incorporation for a corporation.

Of course, in the long term, the team could use conversation-based consensus to ditch basketball and start a knitting circle (though it would potentially “kill” itself in the process, with the same people establishing a new and distinct collective). And the team could use conversation-based consensus to change its decision-making procedure—consensually deciding to appoint a dictator, for example. But such changes take time. Morality cannot demand that a collective change its ways overnight. So the team cannot have an obligation to knit blankets—still less to do so via dictate—within any less time than it’d take to use its current procedure to move toward that. In small groups, this might seem negligible. But in large, complex, and multifaceted groups, such changes can take weeks, months, or even years—as when a state gradually shifts along the political spectrum, or a business gradually shifts into new markets.

4.b Explaining the intuitions

How does Constitutional Constraints explain the intuitions that an art gallery need not give to poverty relief, a small gas station need not overhaul itself, and a state need not extend its franchise abroad? Like Deep Preferences, Constitutional Constraints suggests an error theory. Our intuitions derive from the fact that it looks like a constitutional constraint would be violated. In each case, this appearance is somewhat (but only somewhat) misleading. A Constitutional Constraint may well block the obligation from being performed in the short term. But the relevant timeframe is not short term: obligations to give to relieve poverty, sell sustainable energy, and extend the franchise plausibly operate over the long term. The relevant collectives probably cannot do these things within weeks or even months. But collectives often have diachronic abilities: they are unable to perform some action now, but they are able to take some action now that, if taken, would enable them to perform the action later (Jensen Reference Jensen2009).Footnote ²⁶ If the action would still be valuable at that later time, then the obligation is not ruled out by a constitutional constraint. Our intuitions, then, result from a kind of myopia in which we assume that a present constitutional constraint cannot be changed or that short-term inability implies long-term inability.

Do some collectives have long-term constitutional constraints, which would block even obligations that call for long-term change? Long-term constitutional constraints are exceedingly rare. A long-term constraint via a procedure’s form would imply that some action cannot arise from any possible future procedure that the current procedure could be used to adopt. Decision rules almost always contain rules of change, meaning the procedure’s form is malleable and will almost never amount to a long-term constraint.Footnote ²⁷ Likewise, most collectives are set up to be able to change their ends, meaning they lack long-term fundamental ends (except perhaps the end of being rational). That said, taken literally, constitutions under which a collective’s purpose is “including but not limited to X” imply that the collective really cannot drop X as a purpose (though other purposes can be added).Footnote ²⁸

This might seem too demanding on collectives. If long-term constitutional constraints are rare, and if many obligations concern long-term actions or outcomes, then collectives’ legitimate complaints about obligations are also rare. A less demanding view might say an obligation violates a constitutional constraint only if the collective doesn’t require or expect or usually find that kind of obligation as an input into its procedure, or if the obligation conflicts with its current procedures (rather than procedures it could come to have at a later time when the obligation would still have value), or if the obligation conflicts with its current fundamental ends. Collectives that face these weaker types of “constraints” do seem to be constituted as inept or unreliable regarding the relevant obligations.Footnote ²⁹ Perhaps they are constitutionally incapable of responding to those obligations, even if those obligations make long-term demands.

However, consider two points. First, this permissive standard is not used with humans. Your racist colleague is not let off the hook purely because he doesn’t currently require himself to take seriously his obligation not to be racist, or doesn’t expect that he has obligations not to be racist, or doesn’t usually find the obligation to be racist as a consideration in his decision-making. He’s not let off if his current manner of making decisions only considers the interests of certain racialized groups, or if he is currently unable to let go of racism (as is plausibly the case for implicit biases: these cannot simply be willed away in the short term). These facts do not render him unable to fulfil his obligation not to be racist in the long term. Assuming it would still be valuable for him not to be racist later, once he has changed his racist dispositions, then he has an obligation to do so.

Second, consider our intuitions about wrongful collectives. If it is constitutionally possible for the mafia (understood as criminal) to end its own existence in the long term—the most demanding obligation of all—then we should demand that it do so in the long term. Whether this is constitutionally possible for the mafia is an empirical question not addressed here. I suggest that if we find ourselves resisting such an obligation, then this resistance has arisen because we tacitly believe one or both of two things: first, that the mafia cannot use its decision-making procedures to end its existence (even in the long term); second, that there is little point in asserting the mafia’s obligation because it is unlikely to perform it. The first of these is an empirical issue on which my argument is neutral. The second can easily be debunked: that an entity is unlikely to perform its obligation does not entail that the entity is excused from the obligation or lacks the obligation. So I leave it an open question whether the mafia has this obligation depending on what it is able to do.

If the proposal still seems too demanding, this is likely because some collective obligations are unduly costly on members-qua-private-persons. This provides a supplement to the above error theory. The above error theory suggested that we have overdemandingness intuitions when collectives are unable to perform obligations in the short term, though we err in overlooking that they can perform them in the long term. Additionally, perhaps our intuitions are influenced by the thought that it’s overdemanding on members-qua-private-persons if their collective performs the obligation. As emphasised above, such a thought is not about overdemandingness.

Constitutional Constraints can thereby explain the intuitions from the introduction. What’s more, it shares the other two virtues I attributed to Deep Preferences. First, it respects the result that collectives’ complaints do not rise and fall with the complaints of members-qua-private-persons or members-qua-humans. Constitutional Constraints respects this result because a collective might be constitutionally incapable of producing an outcome within a timeframe, even though its members could produce that outcome in that timeframe by acting ultra vires.

Second, Constitutional Constraints allows for a unified theory across collective and individual agents. Several accounts of individuals’ overdemandingness appeal to psychological inability.Footnote ³⁰ A collective’s constitution can be viewed as its psychological enablers and constrainers: the constitution (combined with inputs) determines what ends up in the collective’s rational point of view. The rational point of view is constituted by (functional equivalents of) mental states. Insofar as a collective’s rational point of view is psychological, the constitution that enables and constrains that point of view generates psychological abilities and inabilities for the collective. So psychological ability is at issue for both individual and collective overdemandingness, if a psychological inability–based theory of individual overdemandingness is correct.

That said, the relevant psychological inability manifests differently across individual versus collective agents. First, individuals do not have the kind of fundamental ends some collectives do. Individuals might have the fundamental end of rationality or abiding by Kantian categorical imperatives (Korsgaard Reference Korsgaard1996), but these are far less domain-specific and task-oriented than “playing basketball.” Second, individuals plausibly lack decision-making procedures like conversation-based consensus, so these won’t produce individuals’ constraints. But this doesn’t mean we can ask more of individuals than of collectives; likely, individuals’ psychological abilities are also restricted in ways collectives’ are not because of individuals’ embodiment and emotion. Psychological inability arises in both types of agent, but it manifests differently.

4.c Underdemanding?

Some might find it underdemanding that constitutional constraints remove some collective obligations. Forget about Mom-and-Pop Gas: Aren’t I letting, say, ExxonMobil off the global-warming hook, at least in the short term? No. ExxonMobil’s diachronic climate-related obligations are plausibly highly demanding. Those obligations will require action in the short term. To show that ExxonMobil can bear a diachronic obligation to move entirely into sustainable energy, we just need to show that its constitutional form and fundamental ends don’t rule out this change in the long run. There are reasons to think such change is within ExxonMobil’s abilities.

First, consider ExxonMobil’s forms. In characterising collectives, I said that informal norms and culture are part of a collective’s decision-making procedure. They are part of the constitutional form. These informal procedures will often be the initial mechanism by which changes occur: water-cooler chat, post-meeting complaining, informal reason-giving, and so on, amongst managers and shareholders. ExxonMobil may be capable of using these informal forms of decision-making to respecify how it enacts its more formal decision-making forms (for example, by revising them to include a wider range of stakeholders).

Second, what about ExxonMobil’s fundamental ends? Isn’t ExxonMobil’s fundamental end “selling oil”? Not necessarily, at least not in the long term. “Supplying energy” may be a more accurate specification of ExxonMobil’s longer-term fundamental end; “selling oil” may be merely a deep preference that is one way of specifying that fundamental end. This is suggested by ExxonMobil’s recent forays into biofuels,Footnote ³¹ and even wind and solar power.Footnote ³² These forays are evidence that, all along, ExxonMobil faced no long-term constitutional constraint against holding sustainable energy as an end. The forays provide retrospective evidence of its always-latent diachronic abilities. If ExxonMobil’s longer-term fundamental end is “suppling energy,” not “drilling oil,” then there’s no reason to think it can’t supply only sustainable energy.

What’s more, even in the rare cases where constitutional constraints let collectives entirely off the hook, other nearby agents will likely have obligations instead. These nearby agents include members-qua-private-persons, different collectives composed of those individuals, individual nonmembers, or collective nonmembers. These agents may have obligations to alter ExxonMobil’s constitution from the outside, to reign ExxonMobil in, or (at the extreme) to “kill” ExxonMobil.

The same goes for other collectives that are rotten to the core. Suppose (simply arguendo) the mafia is indefinitely constitutionally constrained from ceasing to engage in criminality. Suppose, however, that if members act outside their role, for example by ratting, sabotaging, or subverting, then they could bring the mafia down. The individuals are candidates for obligations to perform these actions. If there are several such individuals, they might have obligations to act together, and bear obligations as a collective with an entirely different constitution from that of the mafia itself. Individual nonmembers might have obligations to assist in take-down efforts, as possible. And collective nonmembers (state bodies or other organisations) might have obligations to do what they can. There are, under this scenario, many obligations related to ending rotten organisations. The obligations are simply not held by those organisations (again, assuming there’s a genuine long-term constitutional constraint and assuming obligation implies ability).