Steve Sachs Duke


Tuesday, July 20, 2004


Collective Responsibility: From Thom Ringer:

You've been cited! See footnote #1. Apologies if I have sent this to you already. It's being considered for publication in J.Soc.Pol.Thought.

As I replied to Thom, "Woo hoo! Fame and fortune await."

Thom's very interesting article examines the collective responsibility of group actors, such as corporations or states; it critiques three different approaches before ultimately recommending a fourth. I'll describe the approaches here before adding my own thoughts.

The first view considered is the "mereological view" (MV), which finds its name in what he calls "Sachs' Question" (from a discussion we once had): "Aren't collectives just mereological sums"--the simple aggregates--"of their constituent members?" On this view, the idea of collective action is just a convenient shorthand for the related actions of lots of different individuals. These individuals may have certain expectations, practices, voluntary rules of conduct, etc., and it may be useful sometimes to describe the group as a single actor. But when a collective is said to have acted, nothing occurs over and above a bundle of individual actions. For instance, if we say that "The Army destroyed the bridge," nothing occurred over and above a complex bundle of actions by individual soldiers and officers, which together brought about the bridge's destruction. Thus, when a collective is said to be responsible for a given event, no moral responsibility can be assigned over and above that assigned to various individuals. In the example of the bridge, there are a large number of soldiers who are responsible for helping destroy it, but no separate entity called "The Army" which bears a new kind of collective responsibility.

The second view is the "official conglomerate view" (OCV), which views collective responsibility and individual responsibility as very different things. According to OCV, collectives are groups of individuals that possess "goal-oriented decision-making procedures" which allow them to declare and act upon collective intentions, which are separate from the intentions of the constituent individuals. (For instance, if the Joint Chiefs of Staff decide to destroy the bridge, this desire doesn't need to be shared by all--or even most--of those in the ranks to be considered a collective intention of the military.) According to OCV, collectives can be held responsible for their actions and intentions, in a manner distinct from how we hold individuals responsible for what they do within this framework of collective action.

The third view is the "social wholes view" (SWV), which does not view collectives as groups of individuals at all. Instead, collectives are "non-additive aggregates" of their constituent individuals--the whole is more than the mereological sum of the parts. On this view, a state or corporation is indeed something over and above all the individuals who compose it. It has an independent existence, not merely as a binding framework for coordinated action (as in OCV) or an unstructured bundle of individuals; it adopts a Rousseauian general will that can be entirely distinct from the will of all. The group can act 'on its own,' in a sense, and thus may deserve collective praise or blame for its collective actions.

Finally, Ringer offers, if not a full-fledged account of collective responsibility, a necessary condition for it. A group may properly be described as a collective actor, he claims, only if it is an "epistemic agent" concerning the outcome of its actions. Members of collectives may bear collective responsibility "when they act with a singularity of belief and knowledge about their action such that they can be said to have collectively intended the same outcome." Where such common beliefs are absent, we cannot consider the group to be a true collective; but where they are shared, we can honestly hold the group responsible as a group for its collective actions.


If these issues interest you, I'd encourage you to read Thom's paper, which I've thought about a good deal since he sent it to me. Otherwise, I'd stop reading now, since what follows won't make any sense at all.

Unfortunately, as the putative defender of the mereological view :), I wonder whether the position Thom criticizes isn't actually something of a straw man. To my mind, MV is far more sophisticated and reasonable than it might appear. Moreover, after rejecting the social wholes view and the official conglomerate view, there's not much else left -- not because there aren't any logical alternatives, which there are, but because the motivations behind the rejections of these views don't allow anything short of a full commitment to MV. I'd respectfully argue that the approach Thom suggests in his conclusion is in fact a version of MV, which appreciates some of the benefits that it possesses and that the cruder interpretation may lack. (I've sent a version of what follows to Thom, but thought I'd post it here as well.)

I. The argument against MV

Thom's criticism of MV is that it is "thin" and "inauthentic" as an account of collective action. As an example, he uses Tuomela's case of a full-blown intentional joint action (IJA). In this case, a group of individuals embark on a joint action aware of each other's intentions--and, more importantly, cognizant of this common awareness. Clearly, our moral evaluation of an individual actor

cannot be completely separated from their ... knowledge or speculation about the actions of other participants. To understand the significance of what they did, we have to understand how their intentions were supervened upon by [or themselves supervene on] what they believed or had been led to believe about the actions of other parties. The proponents of MV appear to conflate physical coincidence and epistemic coordination; a superior account, as the examples suggest, will have to say something meaningful about the latter.

I agree fully that an account of collective responsibility must take account of such factors in order to be plausible. Yet I'd also contend that the same condition must be met by plausible accounts of individual responsibility, completely outside of the context of collective action. If Joe purposefully holds down Sam's legs, so that Sam will be unable to escape from a rolling boulder, we can easily assign Joe individual moral responsibility for Sam's death; there is no question of collective responsibility here. Yet if Joe is an innocent bystander who is thrown out of a window and unintentionally falls on Sam's legs, temporarily pinning them to the ground, his moral responsibility is much reduced. The merely physical description of the action is not sufficient to determine our moral evaluation; we need some description of the relevant intentions as well.

Given this background, I see no reason at all why MV cannot, in its moral evaluation of joint actions, take similar account of the intentions, desires, beliefs, states of mind, etc. of those involved. The mere process of "cashing out" collective actions into mereological sums does nothing to eliminate intention. Consider the example Thom provides of "two people raising their glasses" <--> "a toast". Suppose that two people have recently emigrated from a land where the practice of toasting is unknown. After seeing the couples at the tables around them lift their wine glasses, however, the immigrants also raise theirs to the light, in order (as they assume the others have done) to better search for floating insects. This event certainly should not be described as a toast, but the reason has nothing to do with any conglomerate interpretation of toasting. (The idea of a goal-oriented decision-making procedure here would be absurd.) Rather, this is not "a toast" simply because "a toast" requires a common intention to participate in a certain social practice, an intention lacking in the present example. Even after we cash out a supposedly collective action--nothing has occurred here over and above the actions of two individuals--those individuals' intentions and states of mind can still be relevant.

If this last statement is true of mere descriptive accounts (e.g., of a toast), it is even more true of normative accounts. Contra Tuomela, MV has no trouble whatsoever in cases of full-blown intentional joint actions. These would be cashed out into a complex bundle of individual acts, each of which would take place with an intention that depends in a certain way on the intentions of others. (In fact, Tuomela's definition of an IJA constitutes such a cashing-out.) The individual actions of one participant in an IJA could be individually evaluated, if necessary, and this evaluation would quite correctly make reference to their intentions given their beliefs about the actions of others. What makes the collective-action shorthand particularly useful in such situations is that all the parties to an IJA are similarly situated from the standpoint of moral evaluation. We can easily assign moral responsibility to Joe if he purposefully holds down Sam's legs, with the intention that Lucille may then inject Sam with poison. Physically, Joe may have done nothing worse than hold down Sam's legs for a few moments; morally, he quite obviously shares responsibility for the murder, having known in advance what effect his actions would have. The situation is not radically changed if Lucille were aware of Joe's intentions, or if they had discussed the plan for the murder beforehand. Thus, there is no difficulty in asserting that "Joe and Lucille murdered Sam," even though we do not think that anything has occurred beyond the confluence of individual actions, appropriately described. We might think that such pre-meditated, coordinated action is worse than mere awareness of the other's intentions (conspiracy to murder is also a crime), but that does not mean it cannot be cashed out in individual terms. The concept of "collective responsibility" may be convenient in describing such phenomena, especially when the IJA itself is more complicated and involves many participants, but one does not need to invoke a new kind of responsibility or a new kind of entity (such as a collective) to bear it.

More importantly, this understanding of the MV also allows us to explain why we are reluctant to assign moral responsibility to an individual who lacks the requisite intentions, despite his or her participation in a joint act. (I use "intentions" as a catchall phrase, without meaning to require an active will to harm. For instance, if Officer Winky is aware of Joe and Lucille's murder plot, but is indifferent to the outcome and so walks whistling by the scene of the crime, she clearly bears some responsibility for failing to intervene and save Sam's life.) Where these intentions are absent, certain judgments of responsibility should be withheld. MV would not assign responsibility to Joe had he been thrown out of a window and fallen on Sam, pinning his legs to the ground while Lucille injected the poison. Nor would it necessarily assign the same kind of responsibility to Joe if he had been led to believe (with a given level of justification) that Sam was an epileptic who must be restrained during his seizure, and that Lucille was a paramedic injecting Sam with lifesaving medication. (Nor, again, if Joe and Lucille actually were paramedics, and the needles had been switched at the last minute by a nefarious fourth party.) Far from "conflat[ing] physical coincidence and epistemic coordination," MV is perfectly capable of distinguishing the two, and of properly allocating responsibility.

II. The lessons of the alternatives (OCV & SWV)

As I understand it, MV relies on two underlying assumptions:

(1) Ontological individualism: collectives are not ontologically distinct entities from their members. Collectives are constructs. Propositions about collectives are really just shorthand for complicated bundles of propositions about their members.
(2) A Kantian approach to moral evaluation: whether it is appropriate to assign moral praise, blame, or responsibility (of any kind) to individuals depends only on their own voluntary actions, intentions, desires, beliefs, states of mind, etc.

I would infer from Thom's various arguments that he accepts (1) as a metaphysical premise. Certainly it is an appropriate description, as Tuomela notes, of various joint actors such as spontaneous mobs or fluid terrorist cells. (When Qaddafi donated arms to the IRA, he became complicit in their campaigns of terror without becoming the member of any social whole.)

Additionally, I would infer that Thom also accepts (2) in making assignments of responsibility. He criticizes SWV on the grounds that it would overwhelm any relation between responsibility and individual action. (Are babies, or the unconscious, in any way responsible for the current actions of their governments? Yet they make up a part of the social whole.) The criticisms of OCV similarly focus on the unjust assignment of responsibility (good or bad) to those who took no part in the decision-making process for a rule-governed action. The actions of Hugh Thompson, who airlifted civilians at My Lai to safety, would surely seem to exempt him from some degree of collective responsibility. The same could be true of low-level functionaries in Saddam's Iraq who might have collected taxes to support the regime, not merely "under orders" (like Eichmann) but out of personal fear. As Thom notes,

Grudging rulefollowers, particularly those who, in following collective rules, do great harm to others, are morally culpable. But I would argue that they are not morally culpable in the same way as those who enthusiastically followed the collective rules for a single, nefariously focused purpose. It may be that the 'grudgers' are as blameworthy as the 'enthusiastic conformists,' but this is not at all the same as saying that it makes sense to lump both groups together as if they acted collectively and with a singularity of intention.

The only rationale I can see for distinguishing these two cases is that the grudging rulefollowers possess different beliefs or purposes from the enthusiastic participants, and that it would be wrong to hold them responsible in the same way when the requisite intentions are missing.

Of course, these two positions only show that moral evaluation depends in part on intention, not that intentions, states of mind, etc. are the only grounds for assigning responsibility. Yet it seems difficult to explain why some forms of responsibility (call them "individual") may only be assigned given the requisite intentions or beliefs, while others (call them "collective") may be radically independent of our own voluntary actions. What distinguishes the latter types of responsibility from the former? When can an individually blameless person be held collectively responsible?

I can think of only three types of cases where collective responsibility might usefully be distinguished from individual moral evaluation. The first type is collective legal responsibility, such as that of a corporation which faces strict liability for (say) the environmental damage caused by its pollution. This doctrine, however, is quite separate from a doctrine of collective moral responsibility, since legal liability could be assigned, even to blameless individuals, for reasons that have nothing to do with the moral evaluation of a particular case. (For instance, we might want to give the right incentives for due dilligence to future polluters, and thus still hold responsible those who had no scientific reason to fear that the chemicals they emitted might be toxic.)

The second type might be called "corporate" responsibility, where the voluntary members of an institution can be expected to share in the praise or blame it incurs through its coordinated actions. If Coca-Cola violates a contract with a supplier, we think it entirely proper that the money for damages come from the corporate treasury, rather than from the pocket of the person who ordered the contract violation. This is true not only for legal or societal reasons (the incentive effects mentioned above), but also because the institution exists in part in order to be used as a contractual agent. The shareholders, who are the ultimate owners of the corporate treasury, have agreed to bind up their fortunes and entrust them to the managers; they have, in a sense, willingly signed up for a share of responsibility in what comes next. (Such a position might be called a "social-whole-ish" view, given that individuals have agreed to be treated as if they were constituent parts of a given whole.) Even a director who fought the decision before the Board, who campaigned against it at the annual meeting, who has done all an individual could reasonably be expected to do in order to prevent the violation, can have no claim of injustice when, due to the decisions of other shareholders (for which she is blameless), the value of her stock is decreased.

Finally, the third type might be called "unavoidable" corporate responsibility. The model I have in mind here is that of a state, such as the U.S., accepting responsibility for the actions of its officers -- such as by making reparation payments to those forced into Japanese internment camps during World War II. Here again, we think it proper that the money for reparations come from America, from the general taxpayer, rather than from the pocket of those who supported, voted for, or implemented the policy. ("It is America that incarcerated them," one might think, "and America should pay the price.") Yet I would argue that this judgment is more guided by prudence than by morality. Why should someone who emigrated to the U.S. 50 years after the fact, and in ignorance of the World War II-era actions of the government (say, as an illiterate young refugee from genocide), be taxed in order to pay for the misdeeds of prior generations? Assume that it were possible to identify, with perfect certainty, all those who had supported, contributed to, campaigned for, implemented, or just were knowingly indifferent to and accepted the immoral policy; assume further that all of them were still alive, and that they could afford (as a group) the cost of the reparations. Wouldn't it be obviously better to punish only those individuals, according to the degree of their individual responsibility? And wouldn't it be obviously worse to punish, given our perfect knowledge, the young refugee, or the saintly human-rights activist who had taken every morally acceptable measure to prevent the internment camps from being built? (In fact, thinking in terms of moral praise and blame, wouldn't it be just bizarre to place any kind of blame, collective or otherwise, on the refugee? If there is no ontologically distinct entity known as "America" that can be the proper object of blame, then when we say that "America must take the blame," aren't we really just making a complicated assignment of blame to a large number of (perhaps unknown) individual Americans?)

A social-contract theorist might dispute this view, by comparing the state with the corporation in the previous example -- as an institution that one consents, even if only tacitly, to join and defend. (This may be, for instance, why those who take pride in their country -- which no one is forced to do -- sometimes feel as if they should also accept responsibility for its actions. An Iraqi who despised the Ba'athist regime would feel no such compunctions.) But if one does not view these obligations as having been consented to, and if one does not view the nation-state as an ontological social whole, what could possibly be achieved by punishing the blameless in this state of perfect knowledge? Finally, if it is only our lack of knowledge (or the inconvenient death or impoverishment of those responsible) that justifies the use of the general treasury, then this sense that "America must pay the price" does not reveal any new form of responsibility. It is only because of our ignorance and our mortality, not our moral commitments, that the innocent must share the burden.

Thus, unless some additional model for collective responsibility can be found, I can see no reason to saddle the innocent with any form of responsibility that is indifferent to the moral quality of their actions. Whatever Kantian intuitions we might possess against the existence of moral luck should lead us to reject such a notion of collective responsibility which is not reducible to the responsibility of various individuals, or which has not been voluntarily accepted by them. I'm well aware that many people accept the notion of moral luck in theory, and have no difficulty with assigning moral praise or blame to individuals for events entirely beyond their control. Yet without a well-grounded account of why individuals should accept responsibility for particular kinds of moral luck, even when they might have done all they reasonably could to avoid them, assignments of collective guilt to the individually innocent will always appear arbitrary and unjust.

Let's go back to the two claims from the beginning of this section:

(1) Ontological individualism: collectives are not ontologically distinct entities from their members. Collectives are constructs. Propositions about collectives are really just shorthand for complicated bundles of propositions about their members.
(2) A Kantian approach to moral evaluation: whether it is appropriate to assign moral praise, blame, or responsibility (of any kind) to individuals depends only on their own voluntary actions, intentions, desires, beliefs, states of mind, etc.

These assumptions imply that no process of collective organization can impute responsibility to those who are free of such responsibility in their own voluntary actions and intentions. Any attempt to add circumstances under which the innocent can be held responsible, "collectively" or otherwise (e.g., if they are part of a decision-making process, as in the OCV) will fail. A collective is responsible if and only if its members are. As a result, I would contend, these two claims entail a third:

(3) Moral individualism: the only entities that can be assigned moral praise, blame, or responsibility are individuals. Moral statements about collectives are really just shorthand for complicated bundles of such statements about their members. In other words, the mereological view is true.

III. The suggested alternative

At the end of Thom'spaper, he suggests that the key element of collective responsibility is "collective epistemic agency"; that joint actions only entail collective responsibility when the individuals act "with a singularity of belief and knowledge about their action such that they can be said to have collectively intended the same outcome." Thus, the naive chemical engineer, unaware of the deadly uses to which her work will be put, does not share in the same degree of collective responsibility as the rapacious corporate managers who delight in selling weapons to murderous regimes.

Such an account of collective action is very reasonable, but would raise two issues in light of the previous discussion. First, what is the difference between such acts and the full-blown IJAs described by Tuomela? It would seem that if the group were not displaying collective epistemic agency, then it could not be an IJA, for then someone in the group would fail to share the intentions or relevant beliefs of another. And similarly, if the act were not an IJA, it would seem that the condition of collective epistemic agency -- the "singularity of belief and knowledge about their action" -- would be absent. (Perhaps an exception can be found in the case of individuals who intend the same outcome by their actions, but have no knowledge of each other. For instance, suppose there were two independent gunmen on two different grassy knolls, each intending the same outcome and having the exact same beliefs -- such as the belief that "I am the only assassin here." Yet it would seem that the resulting (over-determined) murder is not a case of "collective" action unless they are sufficiently aware of each other's plans; the gunmen might each be guilty of murder, but not of conspiracy thereto.)

Second, if there are no distinctions between acts displaying collective epistemic agency and those that can be described as IJAs, why is the conclusion not merely an endorsement of the mereological view? As argued above, MV can easily accommodate IJAs, because they provide an easy case in which every individual's actions are subject to similar moral evaluation. What is meant by asserting that a group engaged in an IJA has "collective responsibility," over and above the fact that each member already bears a clear individual responsibility for the result? If no one can be collectively responsible without also being individually responsible -- indeed, if the degree of collective responsibility must always be proportionate to that of individual responsibility -- why not accept that the former is merely a convenient means of expressing the latter?

The mereological view is committed to interpreting the actions of collectives as complex bundles of the actions of individuals. In doing so, however, it need not ignore those individuals' beliefs, intentions, desires, or states of mind; indeed, that is why the bundles are often so complex. Accepting such an interpretation on a metaphysical level, and refusing to accept the allocation of praise and blame independent of individual desert, entails a commitment to individualized moral assessment -- to assigning responsibility only to individuals, and not tarring them with their fellows' misdeeds (or praising them for their fellows' wisdom). Our strong intuitive resistance to arbitrary or unjust assignments of responsibility should push us towards a view where individuals are only as responsible as they deserve to be; or, in other words, to the mereological view.




Blog Archives

Front page
XML Feed


© 2011 Stephen E. Sachs


Anglia Regnum