• 1 The Basic Dilemma

  It is not easy to drive a wedge into Philip Pettit’s ingenious and tightly constructed argument, but I am motivated to try because I don’t like his conclusion.
  According to Pettit it is conceptually possible, and at least sometimes desirable, to hold collective agents responsible in a non-metaphorical sense for causing some outcomes, even in cases in which no individual agent can be held fully responsible for causing these outcomes. Collective agents, he argues, may satisfy all the conditions for being held responsible; in particular, they may be autonomous agents, making and executing decisions on the basis of beliefs and desires which can be properly ascribed to them. I will not take any position on this issue but assume, arguendo, that Pettit is right.¹xThe usual argument for collective agency only points out that collectives often can do things, e.g. pass a bill, which none of their constituent members can do. It is a welcome feature of Pettit’s account that he insists that proper collective agents must act for their own reasons. Many collective decision-making bodies, however, do not reason collectively; the members of a majority voting for a bill may all do so for their own individual reasons. And perhaps we should require an autonomous agent not only to act for its own reasons, but also to be able to reflect on the adequacy of its deliberation routines. It may be a problem for Pettit’s account of collective responsibility that only some corporate agents satisfy these conditions; see note 13. But even if collectives can properly be taken to be autonomous agents in their own right, it is an open question whether it is ever a good idea to hold them responsible. Holding collective agents responsible is a practice, and the actual forms that practice takes have to be judged in practical and moral terms, terms of effectiveness and fairness. It is to Pettit’s credit that he fully realizes this. But my objection is that he fails to make that part of his case. It is a moral objection, not an ontological one.
  For clarity’s sake I will focus my discussion on the responsibility of the business firm, the most prominent agent figuring in corporate responsibility case law; in the Netherlands, business firms are thus far the only agent recognized in such case law. The discussion will apply mutatis mutandis to churches, hospitals, universities et cetera. I will also focus on the criminal law. Pettit says explicitly that his argument “makes criminal liability a sensible option to consider with group agents”, but most of his discussion is on a more abstract level. Clearly, however, criminal law is one important social practice of holding agents responsible, as deserving blame and punishment. Pettit insists that there may be good reasons for holding agents responsible without actually imposing on them the sanctions they deserve, but I assume he will agree that holding corporations criminally liable would not be a sensible option if we intended never to punish them. In any case, in holding someone responsible for wrongful harms we at least express blame, and all the points I will make about the allocation of punishment can also be made about the allocation of blame.
  Why is criminal responsibility for collective agents a sensible option according to Pettit? Because it is possible for a group to act wrongly without any of its members acting wrongly or equally wrongly. In that case we can at least punish the group, even if we cannot punish any individuals (to the same extent).
  Such punishments, however, seem objectionable. Suppose, for example, that the company which owned the Herald of Free Enterprise is being held criminally responsible for causing the death of 200 people by negligently allowing its procedures to be as sloppy as they were, and punishment is to be meted out to the company, expressing the blame it deserves. In that case some individuals will in the end have to take on the burdens of that punishment: probably the shareholders; perhaps employees who have to forego pay increases or new job opportunities; or, depending on market conditions, maybe the clients: people who want to travel by ferry from Belgium to England or vice versa. But that is unfair if these people cannot be held individually responsible for the misdemeanor of the company, if they have done nothing for which they personally deserve to be blamed, or at least to be blamed to that extent. That would be obvious if the punishment would involve imprisonment, but it is no less true when it only involves fines, as in all actual cases until now. In such cases the punishment may, on the other hand, still be dissatisfactory to our sense of justice, because a kind of criminal guilt which can normally only be redeemed by the loss of liberty will seem to be bought off.
  It is certainly no less unsatisfactory, if only from the point of view of efficient deterrence, to find that an organization has failed to the extent that this particular company failed, and hold neither the company nor any of its shareholders or staff-members responsible. But do we need to accept that we have no choice other than either of these two infelicitous outcomes: allocating the burdens of punishment to individual people who don’t deserve to be punished, or not punishing any agent at all because no one can be found who is fit to be held responsible? We do not if the law would lay a duty on the people within the company who are in the best position to do so, to change the procedures and the supporting culture of indifference to clients’ interests. Such a law would allow us to hold those same people individually responsible if they fail.²xI agree with Corlett “that the corporate veil of liability created, in the U.S. at least, to shield corporate officials from personal liability for their wrongful deeds qua corporate officials is quite morally unjustified”. See Angelo Corlett, Responsibility and Punishment (Dordrecht: Springer, 2006), 174. But that insight should lead to changes in the law pre-empting the need for what Corlett calls “corporate-collective punishment”. That will require a broader conception of functional or vicarious responsibility than the usual one, a conception which not only looks at direct individual relations between executives and their principals, but also at networks of such relations.³xThis extension has in Dutch law been effected by the Supreme Court judgment of February 23, 1993. See Jaap de Hullu, Materieel Strafrecht (Deventer: Gouda Quint, 2006). I do not deny that such a legal system might include statutory provisions creating corporate responsibility. But we would then be able to distribute that responsibility fairly over individuals, only acknowledging that it depends on the details of actual cases how this is to be done. It would matter, for example, whether the corporate agent has a strongly hierarchical or flat structure of governance. In that case it would turn out, as the Dutch Supreme Court has said in its most relevant judgment on corporate responsibility,⁴xHoge Raad (Dutch Supreme Court) October 21, 2003, specifying criteria for holding corporate agents responsible. Probably the statement also reflects the view that agency cannot be literally ascribed to collectives. that the whole notion is basically a legal fiction: the law could find more cumbersome formulations to avoid it, but yet create the same normative consequences.
  When Eric Colvin raises the question of the fair distribution of the burdens of punishment, he replies that because share-holders stand to profit from the good behavior of their company, it is not unfair if they also have to internalize the costs of its bad behavior.⁵xEric Colvin, “Corporate Personality and Criminal Liability”, Criminal Law Forum 6 (1995): 29. For a similar argument see Peter A. French, Collective and Corporate Responsibility (New York: Columbia University Press, 1984), 188-189. It is only part of their entrepreneurial risk. That is a proper reason, indeed, why they can be held jointly liable for damage caused by the company, but to extend that argument to criminal responsibility means to reduce punishment to a debit entry in the company’s book-keeping, which in no way is understood to express blame. People inside companies may actually think that way – and if a company can reason it may think that way itself – but that makes it no less mistaken. Indeed, we should not reinforce that way of thinking by allowing criminal acts to be only punishable by fines when they have taken place in a corporate framework, when similar acts would otherwise have been punished by imprisonment, and the costs of the fines are thinly spread over people who can easily afford to pay them.
  Colvin goes on to suggest an additional reason why it is not unfair to make shareholders bear some of the costs of corporate negligence: they can take protective action by seeking to have safety systems implemented and by selling their shares if they don’t succeed. But if that is an adequate reply we need no corporate responsibility, for we can allocate criminal responsibility for corporate negligence to the individual shareholders. So, my basic objection to any attempt to answer the accusation of unfairness is that it can only be self-contradictory: if no unfairness would be involved in allocating the burdens of punishment to individuals, we would not need the idea of collective criminal responsibility to begin with, or only need it as a legal fiction, in order to economize on words. Holding a collective agent criminally responsible is necessarily either redundant or unfair: redundant if the responsibility can be distributed without remainder over individual people; unfair if it cannot.

• 2 The Responsibility of Members

  Is it really possible to design the law in such a way that we can fairly factor out the criminal responsibility of a corporate agent over individual agents? Pettit helpfully distinguishes three relevant classes of individual agents: members, designers, and enactors. Focusing on the commercial company, I will also refer to these groups as those of the owners, officials, and executives of the company, meaning by an “official” a person who has authority to regulate the behavior of executives.
  Let me start with the members. The place to start is Pettit’s conception of shared intention, which, as he says, is very close to Bratman’s.⁶xFully developed in Philip Pettit & David Schweikart, “Joint Actions and Group Agents” , Philosophy of the Social Sciences 26 (2006). See Michael Bratman, Faces of Intention (Cambridge: Cambridge University Press, 1999), 98-103. Basically, this conception is the following: A group of two people can be said to have the shared intention to do something – e.g. to sing a duet – if both of its members (i) have an intention of the form: “I intend that we sing this duet”, (ii) have intentions each to sing his part of the score, and (iii) it is common knowledge between them that they have these intentions. If we now look at reason-sensitive collective decision-making as Pettit describes it in section V, we find, however, that this analysis does not apply. In his example, the employees of a firm have to decide about introducing a pay sacrifice. Every individual employee believes that one of the relevant necessary conditions has not been satisfied, but for each condition a majority of the employees believes it to be satisfied. When the employees together decide to go on with the pay sacrifice, it can perhaps be said that all employees intend that “we” pay it, but that is only a consequence of their collective decision. That decision is in no way constituted by the individual intentions of its members, not even their individual “intentions that we …”. That is, indeed, precisely the point of the example.
  Clearly, therefore, we should distinguish between collective intention and shared or joint intention: the collective intention of the incorporated agent, proceeding in accordance with its constitution, and the joint intention of the members of that agent, in particular their intention to incorporate.⁷xPettit is aware of this; see in particular Pettit & Schweikart, “Joint Actions and Group Agents”, 31-33. But he doesn’t explore what it means for the individual responsibilities of members. If we spontaneously decide to sing a duet, there is no collective to which we can ascribe beliefs and desires, and which therefore can have reasons of its own for singing it. Each of us only has his own reasons. Of course, we can form a company for singing duets all over the world, and this company, as an incorporated agent, may have reasons and intentions of its own. But then we are, when singing, playing roles according to the constitution of that agent, we are singing as incorporated executives of that company, and no longer as its unincorporated members.
  It is even true that almost every cooperative activity involves some form of joint decision-making according to some formal or informal rule. If you and I go for a walk together, we will at many places have to decide whether to go right or left, and we do this in ways we know to be acceptable to both of us. So even in that case there is an informal ‘constitution’ and hence a kind of “corporate agent”. In walking we are, so to speak, officials of the walking company. But the original unembedded decision to go for a walk together itself is not a collective decision in that sense. All properly collective intentions of incorporated agents presuppose the merely joint intentions of individual agents to participate in the incorporation, each acting for their own reasons.
  Who exactly belongs to the category of members? Following Herbert Hart, we can say that in any case of incorporated agency, a rule of recognition exists which specifies which other rules are to be recognized as binding on any person who has a role within the corporation, in particular the rules which define the obligations, powers, immunities et cetera of any such role, be it the role of an official or of an executive. The “members” of the agent are the people who are party to its rule of recognition.
  Hart specified that this group should at least consist of the officials of the system. (And in the development of his theory he usually overlooked the possibility that the group could include others as well, as most legal positivists still tend to do.) But an official is a person who has a role which is defined by the rules of the system; therefore, even if the people who are party to the rule of recognition are identical with the officials of the system, that can only be the case contingently. The shared intentions of the members who constitute the system, and the collective intentions of people who have roles in it, should be distinguished, even if these groups overlap.
  The parties to a rule of recognition cannot be said to collectively intend to comply with that rule. The rule is only complied to by individuals for their own reasons, even if these reasons essentially depend on the common knowledge that others will comply with the rule for the same reasons and with the same knowledge.⁸xLegal positivists who also want to be conventionalists have a problem in explaining the rule of recognition as a Lewis-convention because adherence to the rule can hardly be a Nash-equilibrium. That problem cannot be solved by ascribing shared intentions to them on the Bratman-model, however, as Coleman does. See Jules Coleman, The Practice of Principle (Oxford: Oxford University Press, 2001), 94 ff. For such joint intentions still have to be understood in terms of the reasons the individual participants have for joining. The game these people are playing is best understood as an n-persons iterated Prisoner’s Dilemma. Therefore, adherence to a rule prescribing cooperation can hardly be explained without presupposing the adherents to have moral preferences, e.g. for fairness. But this form of conventionalism threatens to undermine positivism. See Govert den Hartogh, Mutual Expectations: a Conventionalist Theory of Law (The Hague: Kluwer, 2002), Chapters 8-9. It is surprising that legal positivists generally fail to address this, the weakest spot in their theory. But if they only have joint intentions, they can only jointly – not collectively – be responsible for the actions which are authorized by that rule. This suggests the following solution to the problem of the allocation of criminal responsibility. The responsibility for whatever is done according to the constitution of the corporate agent, according to its rules, conventional procedures and authorized instructions, can be attributed to its members individually, albeit jointly. It cannot be attributed to them as a collective, because this supposed collective agent has no reasons and hence no intentions of its own. On this view there is no volonte générale, only a volonté de tous.
  If we take this position, we should say about Pettit’s example of collective decision-making that, even if it is true that such a decision cannot be analyzed into the decisions of the officials involved in deliberation, it can in the final analysis be broken down into the individual intentions of the people who together constitute the corporate agent, the owners of the commercial company. They jointly intend everything which is done in their name. It is in this same sense that some people say of a convicted criminal that he is co-responsible for his own conviction.
  In this vein of thinking, the reason why members are individually responsible is only that they are members, that they constitute the corporate agent. They can lay down the responsibility by laying down their membership. (That is why it may be problematic to transfer this line of reasoning to involuntary associations.) However, that one is co-responsible for the existence of an institution⁹xEven this can be doubted in many cases because no single person’s membership is essential to the existence of the institution. Often no one is in a position to make any noticeable difference by exiting. It seems intuitively clear, however, that causal overdetermination does not reduce moral responsibility (the standard example is that of a person shot by two killers at the same time), whatever the proper solution to this problem of moral mathematics, known as the problem of many hands. doesn’t seem enough for being co-responsible for all its operations. A person can only be responsible for the behavior of others if he (at least indirectly) controls it.¹⁰xSince the judgment of the Dutch Supreme Court of February 23, 1954 (IJzerdraad-arrest) this is the most basic condition in Dutch law for both functional (vicarious) and corporate responsibility. It is sometimes suggested that owners do control the behavior of the company, because they have the power to design its constitution.¹¹xKirk Ludwig, “The Argument from Normative Autonomy for Collective Agents”, Journal of Social Philosophy 38 (2007). According to Mäkelä the very fact that a collective agent’s attitudes are not a function of the attitudes of individual decision-makers “is part of the rational design of the collective agent made by the individual members”. See Pekka Mäkelä, “Collective Agents and Moral Responsibility”, Journal of Social Philosophy 38 (2007): 462. But if you control the rules you only partially control behavior according to the rules. Executives may be fully responsible for wrongful harms they cause by their actions, when these have not been required or enjoined by any instructions from their principals, and the principals could not foresee the risks. The relation between members and officials is similar: officials have decision-making powers within the constitution, and therefore have to take at least part of the responsibility for the decisions they make. Otherwise, the members would not need rules for decision-making to begin with; they could start by making decisions for all possible future cases. At the birth of the corporate agent, the members may be fully responsible for its first steps, but the next thing they will usually do is to entrust others (or themselves, but then in another role) with parts of those responsibilities. And in entrusting they transfer not only decision-making powers but also the responsibility for using them.
  There are, to be sure, limitations to the extent to which responsibility can be shifted in this way. If some responsibility has not been farmed out, it should be supposed to remain with the members. If some monitoring has to be done or some decisions have to be made in order to prevent risks to third parties (like the passengers of a ferry), and all the officials of a company can correctly claim that the constitution of the company doesn’t invest them with the power to do this monitoring or to make those decisions (or to change the constitution in this respect), the reproach of negligence can still be laid at the door of the members, in particular when they have been informed about the existence of the danger and the officials’ lack of power to remedy it. In addition, if the basic aim of the company is to make a profit from embezzlement or blackmail, the members cannot wash their hands in innocence and leave it to the officials to bear the burdens of punishment. The very existence of a collective agent with a certain aim, or with a certain constitution, may be seen as involving unacceptable risks to others, and in that case it is still relevant that only the members are responsible for its existence.
  The precise extent of the responsibility of the owners of the company cannot be fixed by such general considerations but only by the specification of the legal duties of the owners. These duties can be more or less extensive, but they cannot be unlimited. In tort law it may be a quite proper construction to create a kind of strict liability for owners: they are always to be held responsible for any damage, and it is up to them to recover the costs of compensation from their negligent or reckless deputies. If they don’t like such risks, they should avoid becoming members. (Harmed parties have no similar options for avoiding being involved.) But it is unfair to impose such strict liability, in the case of criminal responsibility, on people who cannot be held to deserve any blame. Therefore, specifying the duties of owners in law can only be a partial solution to the problem of individualizing responsibility.
  Summarizing: whatever is done by the corporate agent is done on behalf of its members, and if they are not collectively but rather individually responsible for there being an incorporated agent in the first place, it is tempting to conclude that they are not collectively but individually responsible for any action of that corporate agent. But even if the members have the initial responsibility for the actions of that agent, they may entrust others with parts of that responsibility. In that case these delegates can be held responsible for the way they fulfill their task. However, if some responsibility has not been or cannot be shifted, then it is left where it was: with the individual members. We do not want to allow some sorcerer’s apprentices to unleash their little monsters on society with impunity.

• 3 The Responsibility of Officials

  In the last section I argued (1) that, to the extent that the “members” of a collective agent can be held responsible for its actions of the companies, they are necessarily responsible individually rather than collectively, and (2) that the law, by specifying duties, may increase the extent to which they can be held responsible. But I also argued (3) that there is an upper limit of fairness to the extent to which their responsibility can be increased. Therefore I still have to face the most original and most challenging part of Pettit’s argument, which concerns the responsibility of officials for decision-making within the organization, in accordance with its constitution. I have already described that argument. It is possible that all the people involved in decision-making think that a proposal should fail, because at least one of the necessary conditions for accepting it has not been satisfied, but, nevertheless, the decision-making body accepts it because for each of those conditions a majority judges that it has been satisfied. But in that case the decision, and hence the responsibility for it, can only be ascribed to the collective body, not to the officials constituting it.¹²xOddly, Copp uses the alternative procedure for drawing conclusions by taking a vote on the proposal itself to argue that the committee can be blameworthy for concluding that the proposal fails when ‘it’ believes that all the necessary conditions have been satisfied. See David Copp, “The Collective Moral Autonomy Thesis”, Journal of Social Philosophy 38 (2000): 379-380. But in this case it is mistaken (and even inconsistent) to ascribe such beliefs to the committee, only because the majority of its members hold them. Moreover, in this case it is obvious that the committee-members cannot disown the decision which they all agree with.
  How to evaluate this argument? Consider a criminal case which is tried by a single judge. As a private person, he has a strong desire to punish the person who is actually guilty of the crime for which the accused person is being tried, perhaps even a desire to mete out the most severe punishment possible. (You can imagine your own story to make such desires plausible, for example by supposing the victim of the crime to have been the secret mistress of the judge.) And this judge has access to compelling evidence that the accused did commit the crime. Yet, being a conscientious judge, he knows that his private desires are irrelevant to his decision, and it also happens to be the case that the compelling evidence of guilt for some reason is not admissible in court, for example because it has been acquired by the police by means which the law prohibits. So, both the “desires” and the beliefs the judge will use in his deliberation will be completely different from the desires and beliefs he has as a private person; but still it will be the deliberation of an autonomous agent, “autonomous” both in the sense of being sensitive to reasons and of being independent of the reasons of anyone else, in particular of the natural person of the judge. This agent is really an artificial agent, a persona or mask taken on by the natural person. Nevertheless, if we attribute responsibility for his judgment to the judge, we do not mean to suggest that he, the natural person known as Jan Jansen, is not responsible, but only his persona. He cannot disown what he does as a judge. If the law of legal procedure compels him to do what he doesn’t want to be responsible for, for example to convict a clearly innocent person, having exhausted all the (usually ample) space for maneuver the law provides him with, he still has the options of either going beyond his powers or abdicating from the case or from the court.
  Now consider the same case being decided by a court consisting of three judges. This court may decide by votes of 2:1 on the relevant legal provisions, on the allowability of the evidence, and on the meaning of the evidence, yielding the pattern described by Pettit.¹³xPettit suggests that corporate agents are under pressure to design their constitution in ways suitable to prevent inconsistent decisions. For that reason he conceives of the corporate agent as a “novel, consistency-sensitive center of intentional attitude and agency”. (Pettit & Schweikart, “Collective Actions and Group Agents”, see Philip Pettit, “Rationality, Reasoning and Group Agency”, Dialectica 61 (2007). I am somewhat skeptical about the extent to which corporate agents actually satisfy that description. I suspect that many of them do not anticipate the inconsistencies resulting from discursive dilemmas and respond to them in the way most individuals respond to inconsistencies in their plans: by muddling along. It is true that the group decisions at which they will then arrive will still not be a majoritarian function of individual pro-attitudes. But one might question whether such decisions satisfy the necessary conditions for being autonomous. If this is right, the argument for corporate responsibility either needs the empirical premise that corporate agents normally are consistency-sensitive or the normative premise that they can and should be held responsible for ensuring that they are. In that case all the judges will personally disagree with the decision of that court. That court is to a still greater extent an artificial autonomous agent. But still, its decision is the decision of those three judges and none of them can disown it by saying: it is not my judgment but only the judgment of this court. It is true that the judgment of the court is unhooked from their private judgment, but that is no more a reason to disown it then it was in the case of the single judge. As Pettit says, “(t)hey will each intend that together they mimic the performance of a single unified agent”. Hence they are jointly, and not only collectively, responsible for their decision. If they do not want to take that responsibility, they should take appropriate steps, inform the board of commissioners, or the shareholders, or a relevant public monitoring body, or the police, or the press. Or they should step down from the board or leave the company. Unless they act in such ways, they own the decision, and if the decision is criminally negligent, they are. That is why members of a graduation committee sometimes want to leave the committee when they are outvoted (and do so, even if the university rules say they can’t). If there is any doubt about this matter, the law should take it away.¹⁴xNote that this argument does not even require the individual officials involved in the decision procedure to actually, in their role capacity, entertain either the premises or the conclusion of the collective reasoning. For example, the procedure may only require them to take a vote on the criteria, after which a computer announces the decision to the board of directors, of which they are not necessarily even informed. For this reason my argument differs from the one presented by Seumas Miller & Pekka Mäkelä in “The Collectivist Approach to Collective Moral Responsibility”, Metaphilosophy 36 (2005).
  Couldn’t Pettit happily embrace this argument, thereby defusing my moral objection to his proposal?¹⁵xThis is the way in which his paper is understood by Roland Pierik, “Collective Responsibility and National Responsibility”, Critical Review of International Social and Political Philosophy 11 (2008). No, for if the individuals involved in decision-making are jointly responsible for the decisions they make, it does not matter whether together they only mimic the performance of a collective agent or constitute such a performance. It makes no difference whether we see a truly collective agent acting or only a group of people cooperating in making a decision according to an agreed procedure.
  If the members of a board are jointly responsible for the decisions of that board because they cooperated in arriving at those decisions, it doesn’t matter what exactly they did in cooperating, whether they voted for or against any of the judgments on which the final decision rests. An alternative proposal would be to look at the contribution each makes to the final result. If the decision depends on the aggregation of judgments on a number of issues, we should identify the issues on which the collective judgment has been mistaken and excuse the members who rejected that particular judgment, holding the others responsible, not only for that particular collective judgment but also for the final decision which depended on it.¹⁶xSee Matthew Braham & Frank Hindriks, “Judgement Aggregation and Moral Responsibility”, accessible on: <www.rug.nl/staff/m.braham/JA_and_MA.pdf>. In their opinion only normative judgments can be the “guilty” ones, not factual ones. But consider a case in which the judgment that some level of risk is acceptable is correct, but the factual judgment that only a risk of that magnitude is involved is mistaken.
  This proposal, however, neglects the fact that the decision-making officials, when forming their judgments, are already acting in a role-capacity and not as natural persons. We should remember that the format of the decision-procedure Pettit describes is itself the outcome of the constitution of the incorporated agent or of previous decision-making under that constitution.¹⁷xPettit explicitly says that they have “accepted” the format of decision-making in this case, and “agreed” to make the decision in this way. But did they accept it unanimously or by majority, and were they free to choose any other format? Or was it a matter of saying yes or no to “the constitutional formula”? They have to stand for the procedure in exactly the same way as they have to stand for its outcome. For example, it is possible that none of the members of the board agrees that decisions on the three issues which have to be decided in order to arrive at a final decision really are both individually necessary and jointly sufficient. It is even possible that each of them privately thinks that the very premise with which he disagrees is basically irrelevant for the decision; it is only the two others who judged it to be relevant. In that case all three members would fully agree with the final decision, even if each of them rejected one of the premises considered to be essential for arriving at it. But in that case it would be odd to excuse the person who rejected the one mistaken judgment of the three on whom the decision depends. To take another example, all three members may subscribe to Condorcet’s jury theorem or be supporters of deliberative democracy, and therefore interpret the collective judgments as having better epistemic credentials than their own individual view. In that case it would again be odd to excuse them because they have expressed that view.
  So it is not only the final decision of the group which may be different from the decision each individual official would make if he had to make it as a natural person; the voting behavior of each official may be detached in the same way, for example when the procedure only taps him as a source of belief without any regard to the consequences. If, therefore, we could not hold the officials individually responsible for the final decision, we could not hold them responsible for their voting behavior either.¹⁸xPettit cannot appeal to this counter-argument, however, for he believes that individual decision-makers can be excused for the collective decision they make because of their voting behavior: that is why in this kind of case we should hold the group responsible, not its members. See Copp, “The Collective Moral Autonomy Thesis”. Seumas Miller, in “Against the Collective Moral Autonomy Thesis”, Journal of Social Philosophy 38 (2007): 403, similarly believes that it is unreasonable to hold responsible for its adoption committee members who have in vain voted against a certain proposal.

• 4 Conclusion: The Great Chain of Individual Responsibility

  My aim has not been to argue for any particular way of parceling out criminal responsibility among the executives, officials and members of corporate agents. I fully agree that there is no self-evident way. Legal systems will differ in the ways they do it, and these will all have their own pros and cons, including moral pros and cons. My aim has only been to argue that there is an important moral reason of fairness for the law to create a chain of individual criminal responsibility. To the extent that the enactors cannot be held fully responsible, because they followed regulations, conventional procedures and instructions, because they lived in a general culture of indifference to clients’ interests, or because they could not reasonably foresee the effect of their joint actions and omissions, the law should make it possible to identify the officials who are in the best position to design the relevant procedures and to institute ways of monitoring conformity with them, or to take measures, perhaps drastic ones, to change the existing culture. If it turns out that the very constitution of the corporate agent made it impossible for the officials to have acted otherwise than they have done within their powers (e.g. by changing the constitution themselves), responsibility should in the last resort be attributed to its constituting members.
  As I said, one way of creating such a chain of individual responsibility may be a statutory provision (like Art. 51 of the Dutch criminal code) imposing a general duty upon corporations to guard against their operations causing harm and their structures and resources being used to cause it. The law can use that kind of formulation in order to leave it to the corporations themselves to design its own chain of individual responsibility. But that should then be taken to mean that the shareholders will be held responsible if the constitution of the corporation does not allow us in a particular case to clearly identify a particular group of officials as being responsible, e.g. the board of commissioners and/or of directors or a lower level of management. In that case Colvin’s argument applies: they can take their responsibility by insisting on a proper constitution, or sell their shares, if they fail. But that is not an argument in favor of, but rather against, proper corporate responsibility.
  The reason which in Pettit’s own view is the strongest one for imposing a regime of corporate responsibility is that failure to do so creates a perverse incentive.¹⁹xNote that such a regime could also create perverse incentives of its own, e.g. for designing decision procedures which might maximally limit individual responsibility. Members will be stimulated to incorporate in view of arranging things in such a way that nobody can be held fully responsible for harmful outcomes of corporate actions. The individual executives do not know precisely how their joint actions affect the total outcome, and they are under institutional pressure to comply with the rules. The officials don’t have sufficient powers to change the rules. Et cetera. This argument undermines the case for corporate responsibility in the same way in which we found that Colvin’s second reply to the objection of unfairness undermines it. If the owners of a firm conspire to arrange things in such ways, then clearly they can be held individually responsible for the harmful outcomes, in particular if those arrangements explain why officials and executives have not been able to prevent them or can be excused for not preventing them. The same is true if they did not conspire but only failed to see that the arrangements they made would have such consequences. They should have made sure that the arrangements didn’t have such consequences. All the more so when the law explicitly makes them accountable for this.

Noten

• 1 The usual argument for collective agency only points out that collectives often can do things, e.g. pass a bill, which none of their constituent members can do. It is a welcome feature of Pettit’s account that he insists that proper collective agents must act for their own reasons. Many collective decision-making bodies, however, do not reason collectively; the members of a majority voting for a bill may all do so for their own individual reasons. And perhaps we should require an autonomous agent not only to act for its own reasons, but also to be able to reflect on the adequacy of its deliberation routines. It may be a problem for Pettit’s account of collective responsibility that only some corporate agents satisfy these conditions; see note 13.

• 2 I agree with Corlett “that the corporate veil of liability created, in the U.S. at least, to shield corporate officials from personal liability for their wrongful deeds qua corporate officials is quite morally unjustified”. See Angelo Corlett, Responsibility and Punishment (Dordrecht: Springer, 2006), 174. But that insight should lead to changes in the law pre-empting the need for what Corlett calls “corporate-collective punishment”.

• 3 This extension has in Dutch law been effected by the Supreme Court judgment of February 23, 1993. See Jaap de Hullu, Materieel Strafrecht (Deventer: Gouda Quint, 2006).

• 4 Hoge Raad (Dutch Supreme Court) October 21, 2003, specifying criteria for holding corporate agents responsible. Probably the statement also reflects the view that agency cannot be literally ascribed to collectives.

• 5 Eric Colvin, “Corporate Personality and Criminal Liability”, Criminal Law Forum 6 (1995): 29. For a similar argument see Peter A. French, Collective and Corporate Responsibility (New York: Columbia University Press, 1984), 188-189.

• 6 Fully developed in Philip Pettit & David Schweikart, “Joint Actions and Group Agents” , Philosophy of the Social Sciences 26 (2006). See Michael Bratman, Faces of Intention (Cambridge: Cambridge University Press, 1999), 98-103.

• 7 Pettit is aware of this; see in particular Pettit & Schweikart, “Joint Actions and Group Agents”, 31-33. But he doesn’t explore what it means for the individual responsibilities of members.

• 8 Legal positivists who also want to be conventionalists have a problem in explaining the rule of recognition as a Lewis-convention because adherence to the rule can hardly be a Nash-equilibrium. That problem cannot be solved by ascribing shared intentions to them on the Bratman-model, however, as Coleman does. See Jules Coleman, The Practice of Principle (Oxford: Oxford University Press, 2001), 94 ff. For such joint intentions still have to be understood in terms of the reasons the individual participants have for joining. The game these people are playing is best understood as an n-persons iterated Prisoner’s Dilemma. Therefore, adherence to a rule prescribing cooperation can hardly be explained without presupposing the adherents to have moral preferences, e.g. for fairness. But this form of conventionalism threatens to undermine positivism. See Govert den Hartogh, Mutual Expectations: a Conventionalist Theory of Law (The Hague: Kluwer, 2002), Chapters 8-9. It is surprising that legal positivists generally fail to address this, the weakest spot in their theory.

• 9 Even this can be doubted in many cases because no single person’s membership is essential to the existence of the institution. Often no one is in a position to make any noticeable difference by exiting. It seems intuitively clear, however, that causal overdetermination does not reduce moral responsibility (the standard example is that of a person shot by two killers at the same time), whatever the proper solution to this problem of moral mathematics, known as the problem of many hands.

• 10 Since the judgment of the Dutch Supreme Court of February 23, 1954 (IJzerdraad-arrest) this is the most basic condition in Dutch law for both functional (vicarious) and corporate responsibility.

• 11 Kirk Ludwig, “The Argument from Normative Autonomy for Collective Agents”, Journal of Social Philosophy 38 (2007). According to Mäkelä the very fact that a collective agent’s attitudes are not a function of the attitudes of individual decision-makers “is part of the rational design of the collective agent made by the individual members”. See Pekka Mäkelä, “Collective Agents and Moral Responsibility”, Journal of Social Philosophy 38 (2007): 462.

• 12 Oddly, Copp uses the alternative procedure for drawing conclusions by taking a vote on the proposal itself to argue that the committee can be blameworthy for concluding that the proposal fails when ‘it’ believes that all the necessary conditions have been satisfied. See David Copp, “The Collective Moral Autonomy Thesis”, Journal of Social Philosophy 38 (2000): 379-380. But in this case it is mistaken (and even inconsistent) to ascribe such beliefs to the committee, only because the majority of its members hold them. Moreover, in this case it is obvious that the committee-members cannot disown the decision which they all agree with.

• 13 Pettit suggests that corporate agents are under pressure to design their constitution in ways suitable to prevent inconsistent decisions. For that reason he conceives of the corporate agent as a “novel, consistency-sensitive center of intentional attitude and agency”. (Pettit & Schweikart, “Collective Actions and Group Agents”, see Philip Pettit, “Rationality, Reasoning and Group Agency”, Dialectica 61 (2007). I am somewhat skeptical about the extent to which corporate agents actually satisfy that description. I suspect that many of them do not anticipate the inconsistencies resulting from discursive dilemmas and respond to them in the way most individuals respond to inconsistencies in their plans: by muddling along. It is true that the group decisions at which they will then arrive will still not be a majoritarian function of individual pro-attitudes. But one might question whether such decisions satisfy the necessary conditions for being autonomous. If this is right, the argument for corporate responsibility either needs the empirical premise that corporate agents normally are consistency-sensitive or the normative premise that they can and should be held responsible for ensuring that they are.

• 14 Note that this argument does not even require the individual officials involved in the decision procedure to actually, in their role capacity, entertain either the premises or the conclusion of the collective reasoning. For example, the procedure may only require them to take a vote on the criteria, after which a computer announces the decision to the board of directors, of which they are not necessarily even informed. For this reason my argument differs from the one presented by Seumas Miller & Pekka Mäkelä in “The Collectivist Approach to Collective Moral Responsibility”, Metaphilosophy 36 (2005).

• 15 This is the way in which his paper is understood by Roland Pierik, “Collective Responsibility and National Responsibility”, Critical Review of International Social and Political Philosophy 11 (2008).

• 16 See Matthew Braham & Frank Hindriks, “Judgement Aggregation and Moral Responsibility”, accessible on: <www.rug.nl/staff/m.braham/JA_and_MA.pdf>. In their opinion only normative judgments can be the “guilty” ones, not factual ones. But consider a case in which the judgment that some level of risk is acceptable is correct, but the factual judgment that only a risk of that magnitude is involved is mistaken.

• 17 Pettit explicitly says that they have “accepted” the format of decision-making in this case, and “agreed” to make the decision in this way. But did they accept it unanimously or by majority, and were they free to choose any other format? Or was it a matter of saying yes or no to “the constitutional formula”? They have to stand for the procedure in exactly the same way as they have to stand for its outcome.

• 18 Pettit cannot appeal to this counter-argument, however, for he believes that individual decision-makers can be excused for the collective decision they make because of their voting behavior: that is why in this kind of case we should hold the group responsible, not its members. See Copp, “The Collective Moral Autonomy Thesis”. Seumas Miller, in “Against the Collective Moral Autonomy Thesis”, Journal of Social Philosophy 38 (2007): 403, similarly believes that it is unreasonable to hold responsible for its adoption committee members who have in vain voted against a certain proposal.

• 19 Note that such a regime could also create perverse incentives of its own, e.g. for designing decision procedures which might maximally limit individual responsibility.