List and Pettit consider various ways in which groups might adopt procedures that depart from the conditions so as to make possible collective rationality. One possibility is to institute dictatorship, such that the plans of one person are the plans of the group. This is to relax anonymity.54 However, this is in general no solution to the problem of how to enable reasoned joint action, for it achieves coherence at the cost of integration. Alternatively, one can restrict the range of rational individual attitudes that are admissible, which is to relax universal domain.55 However, while deliberation about attitudes may narrow the range of attitudes, it is unlikely to remove the prospect of discursive dilemmas if the set of propositions is sufficiently diverse. It follows that limiting the set of inputs, so to speak, to avoid collective irrationality will curtail the group’s capacity to act on its members’ attitudes.
The most attractive way to maintain collective rationality, suggest List and Pettit, is to liberalize systematicity, which means to take into account how different propositions relate to one another.56 This could be achieved by way of a sequential priority procedure, in which propositions on the agenda are arranged in some order, with earlier propositions taking precedence over later to the extent they are inconsistent.57 Premises might be understood to be earlier than conclusions for example; that is, the premise-based approach is an instance of a sequential priority procedure. The problem with this approach is its rigidity, with much being settled by the initial ordering and hence on how propositions are categorized. This mode of settling which propositions the group adopts is thus somewhat arbitrary and liable to distort joint action. Further, the premise-based approach is difficult to extend to domains in which propositions are not neatly distinguishable into premise and
conclusion.58 Where the agenda is large and complex, and the relationship amongst propositions complex and contested, strict procedures of this kind, while in principle a means to maintain
collective rationality, fail to be an intelligent measure to make possible reasoned joint action.59 The type of group agent these procedures form lacks the openness in response to reason that characterizes well-formed individual agency, in which the relationship between propositions is open to be
revisited, by way of reflective equilibrium.
More promising are procedures that are less rigid and depend instead on making provision for feedback, wherein members of the group are alert to and respond to inconsistency and
incompleteness, drawing out such inconsistency and resolving it. This feedback may be realized by some members of the group privately changing their votes, but more interestingly the group will make provision for inconsistencies to be made public within the group, to be discussed, and then to be resolved to maintain collective rationality (and hence avoid frustrating the objects of the group action).60 For example, the group may use a straw poll procedure, in which propositions are considered in turn and approved on majority vote if consistent with other propositions.61 If
inconsistent, then the group considers (and if need be votes on) the various ways in which one might revise either the later or earlier proposition to ensure consistency. When the members of the group attend to the relations amongst propositions in this kind of way, and strive to settle on a joint position that is complete and consistent, the group is in one sense reasoning about what should be done. One
need not say, as List and Pettit do, that the group takes a view on the relation amongst propositions when it considers possible inconsistencies.62 The group may instead be structured to allow its members to each form a view, such that when they settle on a common plan, that plan is likely to be coherent and reasoned rather than contradictory and arbitrary.
The group that is structured to adopt reasoned plans, which are complete and consistent, forms an agent, which there is reason to think is likely to act rationally. If it acts irrationally, this constitutes a poor exercise of its capacities rather than all that is to be expected from a group incapable of
disciplining itself in response to reasons. The foundation of this group agency is the joint intention to form and maintain a group that responds coherently to reasons. That is, the members of the group act for a shared end and the means to that end, which thus forms an intermediary end in its own right, is to form rational plans that are supported by a stable response to the relevant reasons. The means to that intermediary end consist in a set of dispositions on the part of members of the group and procedures that make it possible for the group to attend to the consistency and completeness of the propositions it adopts. While List and Pettit argue that group agency is possible, even if unlikely, without joint
intention, I disagree.63 Their example—terrorist cells which are coordinated by unknown handlers— seems to me unsound, for the joint intention of the terrorists is to coordinate their actions by way of plans developed by the unknown handlers, who are trusted to pick out means to the shared end that unites the members (say, defeating some enemy).
That group agency is grounded in joint intention is significant. Members of the group aim to act jointly by way of coherent, reasoned plans. Hence they have good reason to adopt and maintain procedures that make inconsistencies salient and provide for their resolution. They also have good reason to change what they would otherwise do, including the option for which they would otherwise vote, in order to maintain group agency. This is implicit in the rationale for the procedures and in their successful application, for when inconsistencies are brought to light, members of the group strive for consistency. Indeed, even in the absence of procedures allowing the relations amongst propositions to be revisited, it would be unreasonable for members of the group to insist on their ex ante plans, ignoring what the group has otherwise decided up to this time. Recall the political party example. Politicians B and C do not act reasonably if they vote to increase other spending, ignoring the party’s earlier decisions (the plan thus far …) not to increase taxes and to increase defence spending. The politicians foresee that their votes will make it the case that the party has an inconsistent platform, which fails to achieve its end, which is to balance the budget and/or to constitute a credible policy on which to contest elections and to govern. Their vote is thus only
rational if one discounts the importance of the joint action in which they are engaged. And indeed this is the point: acting together for some end changes how one should act.
The joint intention to form a reasoned agent, which acts on coherent and consistent plans, changes how members of the group should reason. Consider another example, in this case the members of a company’s board deciding whether to purchase an attractively priced company X, in the specialty chemicals sector.64 The decision involves three propositions.
The company has to decide whether to expand into this sector, whether to do so by acquisition (or by internal growth), and whether to acquire an attractively priced company in that sector. If one decides to expand into this sector and to expand by way of acquisition, then one should acquire X. If the company votes on each proposition, the company decides to expand into this sector, by acquisition rather than internal growth, yet decides not to acquire X. To this extent the company is unable to formulate a reasoned strategy: it adopts reasons that warrant acquiring X, yet fails to acquire X. This is plainly a problem, because it means the company is incapable of executing a coherent strategy, which frustrates profitable action and undermines the company’s credibility as an agent in the
marketplace (it cannot be relied on to respond to reasons or to act consistently over time). Members of the board should grasp this and bring the company’s reasons and action into line. The problem is especially acute if the decisions are made over time, for then the vote by B and C not to acquire X would be to abandon the plan thus far. This is not to say that the plan cannot be revisited, but until it is, it stands as the group’s plan. Hence, when members of the group perceive the importance of
reasoned action they have good reason to discipline themselves and to attend to the coherence of their joint action.
The prospect of collective irrationality is a standing problem for any group that acts on interlinked propositions, especially over time. Many groups cannot attain their defining end without maintaining collective rationality and responding to reasons with complete and consistent joint action. The way to maintain rationality at the group level is first to recognize the problem, to take it to change how one should reason and vote (that is, with an eye to the coherence and intelligibility of the resulting joint action), and to adopt procedures to bring out inconsistencies and to resolve them. When one acts on these procedures, with these dispositions, the group is reasoning and it is well placed for reasoned action. The action of such a group is thus not the mere intersection of private reasoning, for the members of the group are concerned to make it the case that the group acts on a workable plan for sound reasons, and this changes what they have reason to do together.
The intention of the group is not the aggregate of the intentions of its members, for joint action is not the overlap of individual action. Instead, members of some association act for some common end, adopting plans of action that coordinate their joint action. More complex groups adopt complex plans, such that the group has standing intentions and particular intentions. And some groups have good reason to attend to the consistency and completeness of the plans they adopt, such that the group should and may be structured to respond to reasons with coherent, reasoned joint action. Thus, many groups form and act on intentions and some constitute rational agents, which may be expected to act coherently.