Against corporate responsibility

Corporate responsibility is the view that certain groups, and not just their members, can be responsible for their causal impact on the world. And in the moral sense I shall consider, these groups, and not just their members, are seen as appropriate targets of reactive attitudes like resentment and gratitude. A key challenge for defenders of corporate responsibility is to show how a group's actions may be unexplainable strictly in terms of individual actions, and that we instead should attribute the actions to the group itself. They will then have grounds for arguing that there is a class of actions for which the group itself, and not necessarily its members, is responsible. Such an argument will imply that an individualist view leaves some responsibility unaccounted for—we get a responsibility void—and that we need to hold the group itself responsible to avoid a deficit in our ascription of responsibility. Pettit (2003, 2007a) takes on this challenge in his defense of corporate responsibility (see also List & Pettit, 2011: ch. 7). He ascribes responsibility particularly to groups acting on attitudes, or beliefs and desires, formed in a procedure aggregating the group members' attitudes. But when members make consistent attitudes toward a set of logically interconnected propositions, such as {p, p ! q, q}, the majority attitudes might be inconsistent. To avoid inconsistency, the group therefore needs the capacity to adopt attitudes most of its members reject. It can thus form its own irreducible attitudes. And when the group acts on these attitudes, we cannot assign full responsibility to the group members, who do not hold these attitudes. We must instead hold the group itself responsible as an agent in its own right. Otherwise, Pettit warns us, group agents will be let off the hook too easily; “there will be cases where no one is held responsible for actions that are manifestly matters of agential responsibility” (Pettit, 2007a: 197). Received: 30 December 2022 Revised: 26 July 2023 Accepted: 6 August 2023

Here, we should note that Pettit focuses primarily on cases where the issues on the agenda are clearly connected.We shall see that his example of a workers' committee is one such case.In what follows, I also focus on this type of case, and I shall also give little attention to cases where the connection between issues is less clear, and where individuals have a weaker understanding of the consequences of their actions.But it also seems less plausible that individualist explanations leave responsibility voids in such cases.If the problem of seeing how issues are connected is due to poor organizational design, then individuals high up in the organization's hierarchy might be responsible. 1Of course, even well-organized groups can make decisions without a clear understanding of the effects on future decision making.But when individuals' actions have consequences they cannot be reasonably expected to foresee, it seems plausibly that they are not responsible for these consequences (Smith, 1983).But when this happens in collective action, it is unclear why any responsibility would be left unaccounted for.It is unclear why the group, as a distinct moral agent, should have known and therefore can be considered responsible.It seems more plausible to say the group is responsible when individuals can see the connection between issues but are nonetheless said to lack control over how collective decision making on some issue determines a collective decision on some other issue.
In such cases, we might think, with Pettit, that the group, but not its members, is in control.We might think the group is an agent in its own right that satisfies each of Pettit's (2007a: 174) three necessary and jointly sufficient conditions for responsibility.First, it can make a normatively significant choice, which involves the possibility of doing something good or bad, right, or wrong.Second, it has access to relevant information and is therefore able to understand the situation it is in.And third, it possesses necessary control for choosing between options.The choice is truly up to the group.
In this article, I argue against Pettit's account of corporate responsibility by showing why groups cannot satisfy all of these conditions. 2 I especially focus on demonstrating why groups fail to satisfy the third necessary condition for responsibility.To possess the self-control required for meeting this condition, a group's members must not attempt to take control of their group.I show, however that this expectation does not take into account that individuals are agents pursing their own personal ends.We shall see that in cases where inconsistent majority judgments force the group to adopt an attitude the majority rejects, individuals are nonetheless fully responsible due to their capacity to behave differently so as to prevent this outcome.I show in Sections 4 and 5 how the capacity for strategic behavior places the group members in control of their group.And with such control, individuals are fully responsible for their group's behavior.
This demonstration of why groups fail to satisfy the self-control condition for responsibility is the main part of my rejection of corporate responsibility.But I also show why groups might face a problem with satisfying the second condition for responsibility-the access to relevant information condition-that an individual will not face.List and Pettit (2011: 36) understand group members' expressed attitudes to provide their group with relevant information.When we appreciate that these individuals behave strategically, however, we see why they might intentionally feed their group with false information.This weakens the group's ability to understand the situation it is in.

| CORPORATE AGENCY
We commonly talk about groups as being morally responsible for their actions.It might be convenient to refer to a group's intentions and actions, and to say that it is responsible for the consequences of its actions.Nestlé is responsible for polluting a French river, Australia introduced strict travel restrictions during the Covid-19 pandemic, and so on.On closer inspection, however, we might find that such talk is merely metaphorical (Moen, 2023a).The real agents actually responsible are then individuals.
There might also be good strategic reasons for holding a whole group, and not just particular individuals, responsible.Since punishing a group means group members suffer, this practice might give the members an incentive to monitor each other's behavior and to make sure the group functions as the regulator wants it to function.Such a practice does not, however, conflict with an individualist account of moral responsibility (Pettit, 2007a: 200-201).In fact, we can best make sense of it by noticing how it incentivises individuals to behave more responsibly. 3 Arguing that a group is itself morally responsible depends on a demonstration of how it is an agent, and not simply a collection of individual agents.The group must be shown to form and act on attitudes irreducible to those of its members.
Some argue for assigning responsibility to individuals collectively in cases where a specific individual carries out the action, but the action was made possible by others facilitating it or failing to prevent it (May, 1987;Tuomela, 1989).A related case is where an individual performs an action that must be understood in a context created by other individuals.The members of a violent mob, for example, might be considered collectively responsible for collectively "programming" an individual member's harmful action (Shockley, 2007). 4Cases of complicity, similarly, involve the action of one agent made possible by other agents' actions, or inactions.But in these cases, we leave no responsibility void when we hold the individuals involved responsible.List and Pettit (2011) focus on a different kind of case in their defense of corporate agency and responsibility.They propose a model based on groups' autonomous attitude formation.At the basis of this view is the result that any non-dictatorial aggregation function satisfying basic democratic conditions can fail to generate complete and consistent attitudes toward logically interconnected propositions (List & Pettit, 2002). 5The problem is illustrated in Table 1, where each group member submits complete and consistent attitudes toward the propositions on the agenda, but there is nonetheless majority support for the inconsistent attitudes p, p ! q, and ¬q.We must therefore choose between collective rationality and responsiveness to individuals' attitudes-we cannot have both.Pettit (2001a: 272) calls this the "discursive dilemma." Consistent group attitudes are then irreducible to the group members' attitudes.The group must form attitudes rejected by most of the group members.The decision procedure can, for example, assign the majority judgments of p and p ! q to the group, and then go against the majority by assigning q to the group.List and Pettit see this as a cognitive process at the collective level, as the group consciously pursues consistent attitudes.By identifying this as the group's autonomous attitude formation, List and Pettit find no way around treating the group itself as an agent.And if the ascription of agency to groups is indispensable for explaining their behavior, a view of these groups as real agents is warranted (List, 2021(List, : 1215(List, -1217)).Pettit (2003Pettit ( , 2007a) ) takes such a group agent to be responsible, either instead of or in addition to the group members. 6Specifically, he understands it to satisfy the three necessary and jointly sufficient conditions for responsibility mentioned above: it can make normatively significant decisions; it has access to relevant information about the situation it is in; and it possesses the autonomous control needed for choosing between available options (Pettit, 2007a: 174).Anyone satisfying these conditions is, in Pettit's (2007a: 175) view, "a perfectly good candidate for being thought and held responsible."

| CORPORATE RESPONSIBILITY
That a group can satisfy the first condition is uncontroversial since people often make normatively significant collective decisions.List and Pettit (2011: 36) understand groups to also meet the second conditions, the access to relevant information condition, as their members' expressed attitudes are a source of such information.When group members vote in favor of a proposition, p, their group takes that as evidence for p.I return to this condition in Section 7, where I suggest that groups face a problem with accessing relevant information that List and Pettit do not consider.
List and Pettit devote most of their efforts to showing that groups can satisfy also the third condition-the self-control condition.Responsibility depends on a causal connection between the agent's normative judgment and action.This connection is typically difficult to establish, as many factors apart from such judgments determine an agent's behavior.You can form a judgment and intend to act in accordance with it, but your ability to do so will be restricted by your environment.Nonetheless, we usually take individuals to possess sufficient control to qualify as responsible agents.The challenge for a defense of corporate responsibility, Pettit (2007a: 188-189) says, is to show that the problem is no more pressing at the collective level than at the individual level.And List and Pettit (2011: 163) conclude that a group can be "as fit as any individual human being to be held responsible for what it does." They base this view on the aggregation problem described in the previous section, which shows that a group's attitude formation cannot always be responsive to the group members' attitudes.The group may therefore necessarily form attitudes irreducible to those of its members, and List and Pettit (2011: 59) understand this to give the group a "surprising degree of autonomy." In the case Pettit (2001b: 107-110;2003: 183-184;2007a: 197-198) introduces in support of his defense of corporate responsibility, the committee of an employee-owned firm decides on whether to forgo a pay rise to finance new safety measures in the workplace.The three committee members agree that they will forgo the pay rise if and only if the following three conditions hold.
R 1 : The current lack of safety measures entails a serious danger to the workers.R 2 : The new safety measures would effectively improve the workers' safety.R 3 : Forgoing the pay rise would be a bearable cost.R 1 , R 2 , and R 3 are each necessary and jointly sufficient conditions for the decision ((R 1 ^R2 ^R3 ) $ C).
C: Forgo the pay rise.
In Table 2, we see how the committee members' judgments of R 1 , R 2 , and R 3 explain why each of them opposes C. We also see that a majority supports each of R 1 , R 2 , and R 3 , which entails support for C.But the group members unanimously reject C. The majority judgments are therefore inconsistent.The group must therefore either adopt the judgment of C that follows from the majority judgments of R 1 , R 2 , and R 3 , or it must reject at least one of R 1 , R 2 , or R 3 so as to reject C without forming inconsistent judgments.List and Pettit (2011: 56-58) refer to the former solution as the premise-based procedure, and to the latter as the conclusion-based procedure. 7A third solution they discuss is the distance-based procedure, which assigns attitudes based on a "distance metric" (List & Pettit, 2011: 57).Distance here refers to Hamming distance, which compares two strings of binary data in terms of the number of positions that vary across the strings (Pigozzi, 2006).This number is the distance. 8The nearest consistent attitude set, thus understood, is then assigned to the group.In the committee example, this procedure leads to one of R 1 , R 2 , or R 3 being rejected, since that result deviates from only one member's judgment (e.g., A rejecting R 1 ), whereas changing the majority attitude toward C means altering two individual judgments (e.g., A and B accepting C). 9  We see then how the same profile of individuals' attitudes can result in different collective attitudes (Table 3).Under one procedure, the committee decides not to forgo the pay rise, while they make the exact opposite decision under a different procedure.We therefore cannot identify the collective decision to forgo the pay rise at the individual level; it must instead lie at the group level.This is why List and Pettit understand the aggregation problem to force an autonomous group mind into existence.The committee members do not control how the firm acts.If workers complain about not getting a pay rise, List and Pettit (2011: 167) think each committee member can reasonably respond "Don't blame me; I didn't want this result."The agent to blame is instead the committee itself.The committee members might feel bad about their collective decision, but we have no reason to think they did anything wrong (see also Copp, 2006: 215-216).Pettit (2007a: 189-194) acknowledges that individuals both instruct and carry out these actions, and this might make them partly responsible (see also French, 1998: 25).But the group is considered responsible because individuals cannot control the decisions on which these instructions and actions are based.Miller and Makela (2005: 648-649) objects to this defense of corporate responsibility.Whichever priority rule is employed in response to the discursive dilemma, they argue, it will be guided by individuals' votes.They either decide the collective attitudes toward the premises, from which the conclusion is deduced (premise-based procedure), or they decide on the conclusion directly (conclusion-based procedure).Individuals' votes therefore determine the collective attitudes.The discursive dilemma shows that collective attitudes toward some propositions might just be implied by individuals' attitudes toward other propositions.Miller and Makela nonetheless see the process as driven by individuals, since their voting is essential for the outcome.These individuals are therefore responsible for their group's actions.
But while the decision making is based on the group members' voting, as Miller and Makela point out, the discursive dilemma shows why these votes cannot fully determine the collective T A B L E 2 Committee members submit complete and consistent judgments, but the majority judgments are nonetheless inconsistent.attitudes.These attitudes also depend on the procedure employed, which may not have been chosen by the current members, or, in any case, they could not have reasonably foreseen the consequences of the procedure in future cases (Copp, 2006: 217-218;Pettit, 2007a: 187).It therefore seems inevitable that the individuals give up some of the control of their collective decision making.In some cases, no possible outcome will express the attitudes of most of the group members.The group might have to adopt at least one judgment most group members reject.By failing to recognize the group's control over its own attitude formation, Pettit believes we overlook a crucial part of how a group comes to do what it does.By focusing only on individual responsibility, we get a deficit in our ascription of responsibility.

| INDIVIDUAL AGENCY
In this section, I start developing an objection to corporate responsibility that fully accounts for the discursive dilemma.I show why group-agent realism is based on a too restrictive view of individual agency.On a more complete and plausible view, we see that individuals' capacity for strategic behavior gives them control over their group's attitude formation.This strategic behavior becomes evident when we appreciate group members' agenda-setting power and capacity to vote strategically.I shall focus on the premise-based, conclusion-based, and distance-based procedures.These procedures employ a "functionally explicit priority rule," which mechanically assigns attitudes to the group so as to ensure consistency (List, 2004).Procedures with functionally inexplicit priority rules, on the other hand, are more flexible in their ways of solving inconsistencies, and therefore more obviously give control to the group members.In deliberation, for example, the group members give and receive information about each other's attitudes and can then collectively work out how to vote so as to avoid inconsistency (Pettit, 2007b: 512;Pettit, 2009: 81-88;Pettit, 2012: 193-4;Pettit, 2018: 20-21).List and Pettit (2011: 61-64) see the individuals in this procedure as parts of a reflecting group mind trying to make its attitudes consistent (see also Tollefsen, 2002: 401).
However, this procedure is perfectly intelligible from an individualist perspective.In deliberation, individuals work toward the outcome most congruent with their preferences.They may accept attitudes they do not hold themselves as their group's attitudes in order to make sure the group adopts their view on the issue they care the most about (Moen, Forthcoming).The outcome of deliberation is therefore understandable on the basis of the group members' preferences.The case for group agency and corporate responsibility appears stronger when based on procedures with a functionally explicit priority rule, where the group members cannot adjust the priority rule as it suits them.We shall see, however, how individuals control their group's attitude formation also under these procedures.
I shall focus on two ways in which individuals gain such control by manipulating the collective decision making.That is, either by setting the agenda or by voting strategically.Before I consider T A B L E 3 Different procedures deliver different collective judgment sets based on the same profile of individuals' judgments.
these phenomena, I will note that an aggregation procedure is manipulable if and only if a logically possible profile of individuals' attitudes is such that the outcome depends on how the agenda is set (agenda control) or voters can achieve an outcome they consider better from their own perspective by voting untruthfully (strategic voting).Gibbard (1973) and Satterthwaite (1975) show that for preference aggregation, when there are three or more alternatives, any non-dictatorial decision procedure is manipulable.Dietrich and List (2007: 282) demonstrate the analogous result for judgment aggregation that with an agenda of interconnected proposition, no non-dictatorial aggregation function can satisfy the fairness conditions of universal domain and responsiveness as well as collective rationality and non-manipulability. 10Manipulability is therefore ubiquitous, and we shall see why it must be taken into account to explain and assign responsibility for group decisions.

| Agenda control
The individual, or individuals, controlling the agenda determine which alternatives or propositions appear on the agenda, as well as how one might be prioritized over another and the order in which they are voted on.Assuming that agenda setters know other group members' judgments, they can in various ways affect or even determine the outcome.Agenda setters' ability to decide which propositions appear on the agenda can give them considerable influence on the outcome.In the committee example, the agenda setters can put just the outcome on the agenda, thus making sure the employees get their pay rise.This is a conclusion-based procedure.Alternatively, they can put the reasons (R 1 , R 2 , R 3 ) on the agenda to ensure new safety measures and no pay rise.This is a premise-based procedure.In this particular case, this appears unlikely given the committee members' unanimous opposition to this outcome.The point, however, is that agenda setters can decide which procedure is employed, and we see how this decision can effectively determine the outcome.When ascribing responsibility for the outcome, we should therefore consider whether these particular individuals should be given more blame or praise for the group's decision.
We should further note that agenda setters can select items on the agenda so as to construct an outcome of consistent majority judgments.In the example above, reason R 2 is "The new safety measures would effectively improve the workers' safety."But suppose the agenda setters can replace the word "safety" with "productivity" because they expect the committee members to reject this proposition but accept it as a necessary condition for sacrificing the pay rise.The committee will then reject a necessary condition and make a set of judgments consistent with their desired outcome of not forgoing the pay rise.We see, then, another way in which agenda setters affect the outcome in a way for which they are responsible.
So, if we focus only on the inconsistency result and not on how propositions appear on the agenda, we fail to see how the decisions of certain individuals make them particularly responsible for the outcome.We can fill apparent responsibility voids either by arguing that the outcome was constructed by these individuals, or that they should have structured the decisionmaking procedure to avoid a particular outcome.

| Strategic voting
Strategic voting is individuals' response to the way agenda setters structure the decision-making procedure.To see how it works, let us begin by considering the conditions for strategic voting.
First, on a non-simple agenda, an individual, A, must care more about some proposition, or propositions, on the agenda than about other propositions.This is A's "proposition of concern" (List & Pettit, 2011: 112).A second condition is that A's attitudes toward the propositions on the agenda are neither consistently positive nor consistently negative.That is, over the agenda p, q, p ^q, for example, A does not accept all or reject all three propositions.If A's sincere attitudes are consistently positive or consistently negative, she cannot contribute to an inconsistency, and she therefore has no incentive to vote strategically.
The discursive dilemma can arise only if a majority of group members express neither consistently positive nor consistently negative attitudes.If a majority of members hold either consistently positive or consistently negative attitudes, the members' attitudes can be unidimensionally aligned.That is, the individuals can be ordered from left to right on a spectrum such that, for any proposition on the agenda, those accepting the proposition are exclusively to the left, or to the right, of those rejecting it.The collective attitudes will then correspond with those of a median individual (List, 2003).Table 4 presents an example of such alignment.A makes consistently positive judgments, whereas C's judgments are consistently negative (except for the connective proposition, which must be unanimously accepted to make the case relevant).For each proposition, then, the "Yes" judgments are all to the left of the "No" judgments.The majority judgments therefore correspond with the complete and consistent judgments of the median individual, B.
For a discursive dilemma to arise, a majority of group members must therefore hold a combination of positive and negative attitudes.The second condition for strategic voting is therefore satisfied in any such case.Members of this majority can vote strategically by expressing consistently positive or consistently negative attitudes, thus preventing inconsistent majority attitudes that might have resulted in the group not adopting their attitudes on their propositions of concern.
A third condition for strategic voting is that individuals are motivated to misrepresent their attitudes toward some proposition, or propositions, on the agenda.Individuals can be motivated by sincerity-that is, they might be epistemically motivated (Dietrich & List, 2007: 285).To vote strategically, A's desire for collective attitudes she believes to be true must be sufficiently low for her to misrepresent her attitudes so as to ensure that her group adopts her attitude toward her proposition of concern.If A does not hold consistently positive or consistently negative judgments and has no epistemic motivation, then voting strategically is a dominant strategy for A. The greater is A's epistemic motivation, the more even is her concern for all propositions, and the lesser is her incentive for voting insincerely.
On a non-simple agenda, having propositions of concern means having either reasonoriented or conclusion-oriented preferences (Dietrich & List, 2007: 289-291).A has reasonoriented preferences if her proposition of concern is one of the propositions identifiable as a premise, or a reason, for the conclusive proposition on the agenda.A has outcome-oriented preferences if her proposition of concern is the conclusion.In Pettit's committee example, the premises are the three necessary conditions for forgoing the pay rise, R 1 , R 2 , and R 3 , whereas the conclusion is to forgo the pay rise, C.
Individuals with reason-oriented preferences might have an incentive to misrepresent their attitudes toward the conclusion in order to make the majority attitudes consistent with their attitudes toward their propositions of concern.If A's proposition of concern is R 1 or R 2 , she might vote for C to ensure consistency with C.She must also vote for R 3 to ensure that her own judgments are consistent, given "(R 1 ^R2 ^R3 ) $ C." Table 5 shows how the three committee members vote under the conclusion-based procedure if they have reason-oriented preferences.
Given how Pettit presents his committee example, however, it seems more likely that the members have outcome-oriented preferences.They all want to avoid the pay sacrifice but nonetheless end up imposing it, since they employ the premise-based procedure.But with such outcome-oriented preferences, the premise-based procedure incentivises strategic voting.Committee member A, for example, can strategically reject both R 1 and R 2 , which in this case will lead to the majority rejecting R 1 or R 2 , which results in a collective judgment set consistent with rejecting the pay cut (¬C).With outcome-oriented preferences, members B and C also have incentives to submit negative judgments toward the premises they believe to be true in order to ensure that the majority judgments be consistent with rejecting C (Table 6).
Finally, the distance-based procedure gives the members an incentive to misrepresent their beliefs of C insofar as they have reason-based preferences.The result will then be that they accept C as well as all of R 1 , R 2 , and R 3 .If they have outcome-oriented preferences, on the other hand, this procedure gives them an incentive to strategically reject R 1 , R 2 , or R 3 .The result is then a rejection of C and at least one of R 1 , R 2 , and R 3 .On either kind of motivation, strategic voting will lead to consistent majority judgments, and therefore no inconsistency to correct for the distance-based procedure.
We thus see how the committee members' strategic voting leads to an outcome based on their preferences rather than on which procedure is employed (Moen, 2023b).As we see in Table 7, the collective judgment sets are the same under each kind of preference orientation across different procedures.By thus weakening the significance of which procedure is employed, we see how strategic voting can cancel out some of the agenda setters' power (Riker, 1986: 149).
I have so far assumed that the committee members have the same preference orientations in each scenario.It is, of course, also possible that they have different preference orientations, so that some have reason-oriented while others have outcome-oriented preferences.However, strategic voting will nonetheless lead to consistent collective judgments since it involves expressing consistently positive or consistently negative judgments.And that, as we have seen, results in consistent majority judgments corresponding with the judgments of a median individual.
We see, then, how the group members' ability to vote strategically gives them control of their groups in a way List and Pettit do not account for when they argue for groups' autonomy.The group members know that their voting can contribute to an outcome they do not want, and they can act strategically to avoid this outcome.They thus control the collective attitude formation, and no autonomous group agent appears.We have now seen the significant effect of agenda setting and strategic voting on groups' decision making.I now turn to how this individual-level control fills responsibility voids and brings responsibility down to the individual level.The above discussion of agenda control demonstrates how agenda setters can influence collective decisions in ways for which they are responsible.It is less clear how strategic voting implies individual-level control, since a voter usually does not determine an outcome, she only contributes to it.I shall therefore focus primarily on strategic voting here.Braham and van Hees (2011: 10-12) also challenge the idea of individualist responsibility voids on the basis of individuals' capacity for strategic voting.I extend their discussion by demonstrating in more detail how the capacity for strategic thinking makes voters, in addition to agenda setters, responsible in the type of case Pettit focuses on.
A key point to note from the beginning is that recognizing individuals as strategic agents illuminates the need to distinguish between the outcome an individual votes for and the outcome she or he causally contributes to bringing about.The two need not coincide, which is exactly what we see in the discursive dilemma.
Unless an individual is a dictator or a veto player, her contribution to a collective decisionmaking procedure is not a direct choice of a state of affairs.In the cases List, Pettit, and other group-agent realists understand to give rise to responsible group agents, how a group member's choice affects the state of affairs will depend on the other group members' choices.The individual group member thus only has indirect control of the collective decision-making, and her voting can therefore causally contribute to an outcome she has voted against.This is what happens in a discursive dilemma.In Pettit's example, as we have seen, the committee members all reject C, but because two of them accept each of R 1 , R 2 , or R 3 , they contribute to the collective decision in favor of C when the premise-based procedure is employed.They contribute to an outcome no committee member votes for.
T A B L E 5 The committee members vote strategically based on their reason-oriented preferences under the conclusion-based procedure.
The committee members vote strategically on the basis of their outcome-oriented preferences under the premise-based procedure.
The crucial point missed in the defense of corporate agency and responsibility is that we can reasonably expect individuals to know that they can contribute to an outcome they vote against.Hindriks ( 2009) makes a related point in his argument for in-depth analysis of individuals' contribution to a collective outcome.Although his focus is not strategic behavior, he argues that group members should not be let off the hook just because they have not voted for an outcome.When judgment aggregation results in the group making a harmful decision that most of its members reject, the members have made a wrong, or blameworthy, collective decision on a premise.Discussing Pettit's committee example, Hindriks (2009: 173) suggests members B and C are to blame for voting in favor of R 3 , "forgoing the pay rise would be a bearable cost." In my analysis, it is individuals' capacity for strategic thinking that can make them responsible for an outcome they have not voted for.Since they know that their voting on a premise can contribute to an outcome they do not support, they are responsible for this outcome when they in this way do contribute to it.In the committee example, the members know that a vote for R 1 , R 2 , or R 3 could contribute to a group decision in favor of C. Indeed, the members accept that R 1 , R 2 , and R 3 are jointly sufficient conditions for C. So, if their group employs the premisebased procedure, they know that their vote for any of these premises could contribute to their group accepting C, whether they themselves support or reject C.
The committee members also know that strategically rejecting R 1 , R 2 , and R 3 means they cannot contribute to the committee deciding in favor of the pay sacrifice.It is this capacity to think and vote strategically based on their preferences and understanding of the decision procedure that fills the responsibility void that Pettit finds on an individualist account of responsibility.By voting for R 1 , R 2 , or R 3 , the committee members contribute to the inconsistency that may lead the committee to forgo the pay rise.When they choose to vote sincerely, as in Pettit's example, they know that they might contribute to the outcome of forgoing the pay rise.The capacity to vote strategically gives them the option of, with certainty, not being a causal factor for the collective decision to forgo the pay rise.Because of this opportunity, the workers can plausibly blame the committee members when they do not receive the pay rise.When the committee members point out that they voted against this outcome, the workers can nonetheless meaningfully hold them responsible for it.After all, each of them voted in a way they knew, or at least should have known, could contribute to this outcome despite having the opportunity to vote in a way that could not possibly contribute to this outcome.
To further clarify this point, let us consider another case where individuals are responsible for an outcome they did not vote for but nonetheless contributed to. 11Most Americans whose first preference for president in the 2000 election was Ralph Nader probably had Al Gore as their second preference.They also knew Gore had a far better chance of winning than Nader had.Given the plurality system, their sincere voting therefore helped their third preference, George W. Bush, win the election. 12We should not think of these voters as innocently expressing their preferences and then blame, or praise, the electoral system for making Bush T A B L E 7 Individuals' preferences, and not the decision procedure, determine the collective judgments.

Reason-oriented preferences
Outcome-oriented preferences the winner.Doing so would be to underestimate individuals' ability to understand and respond to their social environment.Now, Pettit's committee members might have strong epistemic motivations outweighing any outcome-oriented preferences.They will then vote sincerely to produce the outcome of Pettit's example.But they also know that doing so might contribute to an outcome they do not support.Voting takes place in a strategic environment, where what one ought to do depends on how one's actions will combine with the actions of others to produce an outcome.Sincerity is therefore a choice for which the committee members are responsible.The committee members should be ready to take responsibility for their sincere voting.That a discursive dilemma occurs is therefore no reason for assigning responsibility to the group instead of its members.

| LIMITS TO STRATEGIC BEHAVIOR
In some cases, of course, people will not vote strategically.List and Pettit (2011: 113-114, 124-128) discuss two ways of making individuals express their judgments truthfully.The first is to take their preference orientations as fixed, and then employ a procedure that does not incentivize strategic voting.As we have seen, if the individuals have reason-oriented preferences, the premise-based procedure gives them no incentive to vote untruthfully, and if they have outcome-oriented preferences, the conclusion-based procedure makes truthfulness incentivecompatible (Dietrich & List, 2007: 289-291).Choosing a procedure based on preference orientation can thus disincentivise strategic voting (List & Pettit, 2011: 113).But this is no way of reducing individual control of collective attitude formation, as the attitudes will still reflect individuals' preferences.In light out the above discussion, we can also see this way of incentivizing truthfulness as the agenda setter's construction.So, individuals remain in control and are responsible for the outcome.List and Pettit's (2011: 113) other solution is to try to change the individuals' preferences by convincing them that they should care about a different set of propositions of concern.By making everyone's preferences reason-oriented, we can then employ the premise-based procedure.Alternatively, by making everyone's preferences outcome-oriented, we can employ the conclusion-based procedure.Again, however, individuals' preferences will determine the outcome, though they might here share the responsibility with whoever is cleverly making them adopt a particular preference orientation.
A more interesting version of this second "behavioral" solution is to make individuals adopt no particular preference orientation but a preference for honesty.That is, a preference for expressing attitudes truthfully on all propositions as if they were separate and not inter-connected.Most likely, this is the case where individuals internalize the conviction that strategic voting is morally impermissible.Braham and van Hees (2011) discuss one such example.While they are largely dismissive of the possibility of responsibility voids based on the observation that individuals need not submit sincere judgments, they concede that members of the jury in a court trial might be morally obligated to express sincere judgments without regard for the outcome.
It is not clear, however, why a responsibility void would appear even in such a case.We may suppose the jury members fulfill their duty to vote sincerely, but it still does not follow that a collective-level inconsistency will result in a responsibility void to be filled by a corporate agent.We should also account for the responsibility of the individuals deciding how to handle an inconsistency.They should perhaps be required to make the outcome proposition (e.g., "the defendant is guilty") the most important proposition, and therefore employ the conclusion-based procedure.Or they might be morally required to make it a rule that the defendant is guilty if and only if both the premise-based and conclusion-based procedures produce a guilty verdict.Inconsistent majority judgments would then result in the defendant's acquittal.
We might also think groups handling issues so serious that the members are morally obligated to vote sincerely should employ a super-majority rule.With k propositions, a super-majority greater than (k -1)/k is guaranteed to produce consistent judgments (List & Pettit, 2002: 106).With three propositions, then, a super-majority larger than two-thirds will ensure consistency.With four propositions, a super-majority of more than three-quarters ensure consistency, and so on.With such a rule, members will have no incentive to vote insincerely.The point here is that by declaring a responsibility void, we overlook the moral significance of institutional design.
By appreciating how individuals are strategic agents, then, we see why the discursive dilemma is no basis for the idea of a responsible corporate agent.We can distribute all the responsibility across the group members, which means no responsibility is left for an irreducible corporate agent.The responsibility void Pettit finds is not there.
A separate question is how responsibility should be distributed in discursive-dilemma cases.In the committee case, no one member seems more responsible than another, since each accepts two reasons and rejects the conclusion.In other cases, however, we might find that responsibility should be distributed unequally.Suppose a three-member group makes judgments on three propositions, where p and q are premises that jointly imply r.Member A supports all the propositions, while B and C support only a premise each and reject the conclusion.With the premise-based procedure, we get a judgment of the conclusion that B and C, who constitute a majority, reject (Table 8).Here, we can plausibly judge A to be more responsible than B and C, since A voted directly for the outcome r.But B and C will nonetheless have some part of the responsibility because they could have prevented r by voting against p and q.The relevant difference between A, on the one hand, and B and C, on the other, is that A's expressed attitudes can contribute to no other outcome than r, while B and C's expressed attitudes can contribute to either r or ¬r.So, to avoid responsibility for the collective decision in favor of r, B and C should express the attitude set {¬p, ¬q, ¬r}, which can only contribute to ¬r.

| ACCESS TO EVIDENCE
I have so far only examined Pettit's third necessary condition for responsibility concerning groups' self-control.But I shall now also briefly question groups' ability to satisfy Pettit's second necessary condition for responsibility.This is the condition that the agent has access to relevant information and can therefore understand its situation and see what is at stake.List and Pettit (2011: 36) understand group members as a group's "eyes and ears."When they vote in favor of a proposition, p, the group takes that as evidence for p.The group members' votes are therefore to the group what perceptual evidence is to an individual.
One potential problem with this view appears when we consider how group members might strategically misrepresent their beliefs.They can thus intentionally deny their group relevant information.Here, I do not mean to challenge the common observation that by aggregating individuals' knowledge on an issue, the group can form a view more accurate than that of most, and perhaps even all, of the group members. 13The point, however, is that individuals' strategic behavior can produce a problem for belief formation at the group level that does not arise at the individual level.
This problem is best illustrated by a case in which the group members vote for a proposition they all believe to be false, or against a proposition they believe to be true.This happens in the committee case when the conclusion-based procedure is employed, and the group members have reason-oriented preferences.The group members then vote in favor of C despite all being opposed to C. We thus see how strategic voting can lead to a collective judgment in favor of a proposition all the group members reject (Table 5). 14 Now, "forgo the pay rise" is not a proposition that is clearly true or false as a matter of fact.This is evident in the group members' voting, as two of them sincerely accept each of the propositions that are necessary and jointly sufficient for C. A clearer example of group members strategically denying their group relevant information would involve a proposition that all group members correctly believe to be true (or false) and is not entailed by other propositions.Suppose the three members of a political party vote to decide the party's stand on a mandatory vaccination program.And to justify the party's stance, they also vote on three propositions, R 1 , R 2 , and R 3 that are jointly sufficient reasons for C. R 1 : The vaccine will protect the population against a serious disease.9, the premise-based procedure will produce the decision on C that a majority of the group members (B and C) prefer.Without complete information of how the other members will vote on each proposition, however, B and C have an incentive to vote against each reason on the agenda (¬R 1 , ¬R 2 , ¬R 3 ) to ensure that the premisebased procedure will lead to the rejection of C. Members B and C therefore vote strategically against R 1 , thus ensuring that the group rejects R 1 despite all the group members believing it to be true.
It would, of course, be too demanding to take responsibility to require attitudes consistently congruent with available evidence.On this condition, no individual would pass as responsible.But we can look for explanations for an individual's false attitudes that reveal why the individual is nonetheless responsible.For example, the individual may have decided that gathering more available information was too costly, which is a decision the individual can plausibly be responsible for.But such explanations are impossible when a group forms false judgments as a result of its members' strategically misrepresenting their beliefs.Here, the group's lack of access to relevant information is not the group's own decision but the result of its members' decisions.The responsibility, then, is with these individuals.Individuals' strategic behavior can thus weaken their group's ability to satisfy the access to relevant information condition for responsibility.

| CONCLUSION
Based on rigorous social choice theory, List and Pettit show how no decision procedure can be perfectly responsive to the group members' attitudes.For the sake of consistency, it might have to assign attitudes to the group that most, or even all, of the group members reject.On a strictly individualist account, Pettit believes we fail to account for the responsibility of behavior based on such irreducible collective attitudes.We must therefore ascribe responsibility to the group in order to avoid leaving some responsibility unaccounted for.In this article, I have argued against this defense of corporate responsibility by showing how an individualist view can fully account for the responsibility of a group's behavior.I have done so by demonstrating the significance of agenda control and strategic voting.Agenda setters can influence and perhaps even determine the outcome by selecting the items on the agenda and the procedure for how to vote on them.And by recognizing that individuals can vote strategically, we see how the collective-level inconsistency that corporate responsibility depends on is a result of individuals contributing to an outcome they have not voted for, as well as their capacity to vote so as to avoid contributing to it.On a complete account of individual agency, we see that the group members understand that the choices they make within the group structure can lead to an outcome they do not support.While no voter may be able to directly determine the outcome, the voters know how to vote so as to be sure not to contribute a certain result.Group members therefore cannot be said to leave their group in control of the attitude formation.The group does not possess the self-control Pettit takes to be necessary for responsible agency.To claim that it does is to underestimate individual agency.
We have also seen how groups might also face a problem meeting another of Pettit's necessary conditions for responsibility, or at least unable to satisfy it to the same extent as individuals can.When group members vote strategically, they provide their group with information they believe to be false.They thus deny their group access to relevant information and make it ill-equipped for making decisions for which it is responsible.
Pettit sets out to show that problems of ascribing responsibility to groups are no greater than the problems of regarding individuals as responsible for their actions.And with List, he concludes that it is a mistake to think group members deprive their group of the control required for responsibility, just as it is wrong to think neurons deprive an individual of such control (List & Pettit, 2011: 161).By fully appreciating that groups are themselves made up of strategic T A B L E 9 A political party rejects a proposition, R 1 , all its members believe to be true.agents, however, we see why a defense of corporate responsibility faces considerable problems not present at the individual level.We see how individuals can control their group, and why they are therefore fully responsible for its behavior.

R 2 :
The mandatory vaccination program would involve a tolerable infringement on individual liberty.R 3 : Financing the mandatory vaccination program would be a justified use of public funds.(R 1 ^R2 ^R3 ) ! C: If R 1 , R 2 , and R 3 , then the party should support the mandatory vaccination program.C: The party should support the mandatory vaccination program.The party employs the reason-based procedure, and the party members have outcomeoriented preferences.As we can see in Table Each group member expresses consistent attitudes toward the propositions on the agenda, but the majority attitudes are nonetheless inconsistent.
Unidimensional alignment due to A's consistently positive judgments and C's consistently negative judgments resulting in majority judgments corresponding with B's judgments.
T A B L E 8 Under the premise-based procedure, member A contributes directly to the collective attitude in support of r, while B and C contribute indirectly to r.