An Easy Case: States

I’ll assume that, in the most general and abstract terms, a collective agent is composed of (at least two) human agents that are united under a rationally operated group-level decision-making procedure, while non-agential groups are composed of human agents that are not united in this way.

I’ll assume there are three jointly sufficient conditions for a decision-making procedure to be group-level and for individuals to be united under that procedure. When these three conditions are met, the individuals who are united are the collective’s members. So I’ll call them the “members” in what follows.

First, each member is committed (even if only tacitly) to abide by the procedure’s results. This condition is important for enabling the collective to act in the world. After all, members are the physical constituents of collectives, so the collective can physically act only if it does so through the physical form of its members. But in order for the collective to reliably control these physical constituents, the constituents must be committed to enacting the collective’s decisions. This commitment is pro tanto—it can be overridden—but the commitment is pre­sumptively decisive in the individual’s decision-making.

Second, the beliefs and desires that the collective’s procedure takes as inputs—and the way the procedure processes those inputs to form decisions—systematically depend on the deference, votes, debate, contributions, etc., of its members, while being operationally distinct from the inputs and processes that any member uses when deciding for herself. This condition is important for locating the collective’s agency as distinct from its members’ agency. The collective’s inputs and processes are operationally distinct insofar as: first, the beliefs and desires that are the outputs of the group-level procedure can have different content than the beliefs and desires that any of the members input into the procedure; second, its method for processing the beliefs and desires (that are produced by the procedure) is different from the method of any one member when deciding for herself; and third, the decisions it produces are not the straightforward conjunction of individuals’ decisions.

Take the UK again. The beliefs, desires, and decisions of the UK systematically depend on the manifestos of political parties, the decisions of ministers, the judgments of judges and juries, and so on. The first two of these, at least, in turn systematically depend on the votes, letters, petitions, and so on, of citizens (and, sometimes, some non-citizen residents). So the UK’s decisions systematically depend on those of its component members (though, again, citizens are not component members if they don’t meet the commitment condition). Although this sys­tematic dependence holds, the UK’s beliefs, desires, and decisions are operationally distinct from those of its members, in the three ways outlined above: first, the UK can believe something (e.g. “a bespoke Brexit deal is possible”) even while many, most, or even (in theory) none of the members personally believe this; second, the UK’s methods for using its beliefs and desires to form decisions include parliamentary debates, Cabinet meetings, and so on—these are decidedly not the procedures that members use when using their own beliefs and desires to make their personal decisions; and, finally, the decisions of the UK are not just “Theresa May decides this and Boris Johnson decides that and....” Instead, the UK resolves such disagreements to arrive at an unequivocal collective position (this is most clearly exemplified in the “collective respon­sibility” of Cabinet (Brady 1999)).

The third jointly-sufficient condition on a decision-making procedure’s being group-level, and of individuals being united under the procedure, is that the enactment of the group’s decisions requires actions on the part of the members, where those actions are properly under­stood as attributable to the collective. This third condition will hold if there is a particular inter­play between the first two conditions: not only are individuals committed (so that the collective can, by and large, determine their actions-qua-members), and not only does the collective make decisions in a distinctive way, but the distinctively collective decisions and the actions of the committed individuals converge to make some actions of members attributable to the collective as such.

To illustrate, suppose the UK decides that it will teach its schoolchildren a particular curriculum. The enactment of this decision requires actions on the part of thousands of members: teachers, regulators, curriculum designers, and so on. The actions of a given teacher, when teaching this curriculum, are his or her own. But they are also properly attributable to the state, because they result from his or her commitment to abide by the state’s distinctive decision­making procedure. To see this, we can imagine that the state takes the position that evolutionary theory will be taught in biology classes. To a teacher who doesn’t believe evolutionary theory, his or her action of teaching evolutionary theory is his or her own—but it cannot be properly understood as due to his or her agency alone. He or she is acting as an organ, part, participant, or (I will usually say) member of the state. Without referring to the state, we cannot locate one of the decision-makers (alongside himself or herself) that is behind his or her action, since the group’s decisions explain his or her action and he or she would not perform that action if acting in a non-member capacity.

As it is hopefully clear from the discussion of the UK, all states have collective agency. Of course, not all states have ordinary citizens as members (in the technical sense of membership at issue here): dictatorships and aristocracies, for example, will include a much narrower group of humans as members that meet the three conditions just outlined.

The idea that states are agents accords with the views of prominent international relations theorists (e.g., Erskine 2001; O’Neill 1986; Wendt 2004), and with the long tradition of realism in international relations theory. According to realism, states are rational and self­interested agents, who pursue their own power, preservation, and self-interest in perpetual tension with other states. Realism assumes that states are agents, but it does not impute respon­sibility to states, because responsibility requires moral agency.

Do states have moral agency? Moral agents are agents that can deliberate about, and act on, morally good reasons. Such deliberation and action plays a crucial role in prospective respon­sibility (an entity must be able to deliberate and act upon morally good reason if it is to fulfill its obligations) and in retrospective responsibility (we typically blame and praise only those entities that can deliberate and act upon morally good reasons). There is no conceptual road­block to a collective being a moral agent. After all, human moral agents can recognize the morally good reasons that apply to agents other than themselves. If a collective’s members are human moral agents, and so can recognize such reasons, then they will be able to (i) design a collective’s decision-making procedure such that the procedure countenances the input of the morally good reasons that bear on the collective’s decision-making, and then (ii) operate the procedure so it processes those reasons in the way morality demands of the collective, such that the collective forms its own intentions to act in response to those reasons. Of course, it’s unlikely that any collective will respond to morally good reasons infallibly, but if infallibility were the standard for moral agency, then no human would be a moral agent. Our conception of moral agency should not require moral perfection, so this is not a conceptual roadblock to collectives’ moral responsibility.

However, if a collective is constitutionally incapable (not in terms of composition, but rather in terms of governmental capacity) of taking morally good reasons as inputs and processing them, then the collective cannot be held morally responsible: it can neither bear obligations nor be blameworthy.

If members are not permitted (by the procedure) to present morally good reasons to the col­lective, then their attempts to do so will result in “stonewalling” by the collective. The collective is then the moral equivalent of a psychopath: rational, but amoral (Hindriks, forthcoming, also draws this analogy. The Nazi Party might be an example of such a collective). To picture this, imagine sitting on a committee and raising a consideration outside the committee’s remit. While you have tried to present this consideration to the committee, the committee’s constitution does not permit it to be considered—so the committee is constitutionally incapable of acting on that consideration. If the procedure does not permit morally good reasons to serve as inputs to the collective’s deliberation, then the collective will not be able to act on such reasons. But if mor­ally good reasons can be input into the collective’s procedure, and if those inputs are in fact put into the procedure, then the collective will deliberate upon them—using such mechanisms as inter-member debate, member reflection, established collective practice, and so on. Such delib­eration is not algorithmic or automatic: a collective’s use of its procedures to deliberate upon morally good reasons is thus similar to the deliberation of human moral agents. It can be held prospectively and retrospectively responsible for the results of its deliberation.

How can a state have moral agency? By having fora in which members are permitted (by the group) to input morally good reasons into the group’s decisions and within which delib­eration on such reasons occurs. Examples of such fora include petitions and protests that are then discussed in the legislature; parliamentary debate in general; and election campaigns with subsequent voting (the slogans for which regularly appeal to citizens’ moral sensibilities). By having such fora, a state is not just a collective agent, but also a moral collective agent. That said, states need not be democratic to be moral agents: conditions (i) and (ii) for moral agency are multiply realizable. Even a dictatorship or aristocracy can be a moral agent—just not one of which ordinary citizens are members.

However, it’s possible that not all states are moral agents. This is because—arguably—not all states have decision-making procedures that permit their members to put morally good con­siderations as inputs into the procedure. Swaziland or the DPRK are perhaps examples of states that are agents. After all, even in an absolute monarchy—where the state has just one member, in the sense outlined above—that member’s decision-making processes will differ when he is acting qua head-of-state than when he is acting qua individual. But such states are not moral agents, that is, capable of acting on morally good considerations or reasons. If a group’s decision­making procedure has immoral goals as its absolute priority (the goal of maintaining the power of the leader, say), then feeding morally good inputs into the procedure may not be permitted by that procedure. Another way to say this is that morally good considerations are ruled out by the state’s constitution: the state is a psychopath.²

Crucially, this is consistent with thinking that other entities—whether individual or collective—have obligations to make that state into a moral agent. For example, dictators may well have obligations to do so. After all, states’ leaders are moral agents—constitutionally capable of having morally good considerations bear on their decisions. So, they are capable of bearing responsibility for the actions performed by the states they control. Even if the state’s leader is a psychopath, other agents can bear obligations to act upon the state and its leader to render the state a moral agent. To be clear, I have not here argued that such obligations always exist. I just mean to emphasize that such individually-held obligations (and blameworthiness for not dis­charging them) are not ruled out by the fact that the relevant state is not itself a moral agent. If such individually-held obligations exist, then individuals can (and, often, should) be held blame­worthy and culpable when the non-moral-agent state causes harm.

23.4