Co-responsibility and Causal Involvement

Abstract

In discussions of moral responsibility for collectively produced effects, it is not uncommon to assume that we have to abandon the view that causal involvement is a necessary condition for individual co-responsibility. In general, considerations of cases where there is “a mismatch between the wrong a group commits and the apparent causal contributions for which we can hold individuals responsible” motivate this move. According to Brian Lawson, “solving this problem requires an approach that deemphasizes the importance of causal contributions”. Christopher Kutz’s theory of complicitious accountability in Complicity from 2000 is probably the most well-known approach of that kind. Standard examples are supposed to illustrate mismatches of three different kinds: an agent may be morally co-responsible for an event to a high degree even if her causal contribution to that event is a) very small, b) imperceptible, or c) non-existent (in overdetermination cases). From such examples, Kutz and others conclude that principles of complicitious accountability cannot include a condition of causal involvement. In the present paper, I defend the causal involvement condition for co-responsibility. These are my lines of argument: First, overdetermination cases can be accommodated within a theory of coresponsibility without giving up the causality condition. Kutz and others oversimplify the relation between counterfactual dependence and causation, and they overlook the possibility that causal relations other than marginal contribution could be morally relevant. Second, harmful effects are sometimes overdetermined by non-collective sets of acts. Over-farming, or the greenhouse effect, might be cases of that kind. In such cases, there need not be any formal organization, any unifying intentions, or any other noncausal criterion of membership available. If we give up the causal condition for coresponsibility it will be impossible to delimit the morally relevant set of acts related to those harms. Since we sometimes find it fair to blame people for such harms, we must question the argument from overdetermination. Third, although problems about imperceptible effects or aggregation of very small effects are morally important, e.g. when we consider degrees of blameworthiness or epistemic limitations in reasoning about how to assign responsibility for specific harms, they are irrelevant to the issue of whether causal involvement is necessary for complicity. Fourth, the costs of rejecting the causality condition for complicity are high. Causation is an explicit and essential element in most doctrines of legal liability and it is central in common sense views of moral responsibility. Giving up this condition could have radical and unwanted consequences for legal security and predictability. However, it is not only for pragmatic reasons and because it is a default position that we should require stronger arguments (than conflicting intuitions about “mismatches”) before giving up the causality condition. An essential element in holding someone to account for an event is the assumption that her actions and intentions are part of the explanation of why that event occurred. If we give up that element, it is difficult to see which important function responsibility assignments could have.

Introduction

In discussions of moral responsibility for collectively produced effects, some reject the view that causal involvement is a necessary condition for individual co-responsibility. In general, considerations of cases where there is “a mismatch between the wrong a group commits and the apparent causal contributions for which we can hold individuals responsible” motivate this move.¹ According to Brian Lawson, “solving this problem requires an approach that deemphasizes the importance of causal contributions”.² Christopher Kutz’s theory of complicitious accountability in Complicity from 2000 is probably the most well-known approach of that kind.

Consider the Allied strategic bombing of Dresden 1945. Kutz describes the event as displaying the following features. The bomb raids were strategically valueless and inhumane, and this should have been “and almost certainly was” apparent when the decision was made.³ Each of the 8000 crewmen’s causal contribution was “imperceptible”⁴ and “marginal to the point of insignificance”.⁵ Furthermore, the overall effect was over-determined. “No rank-and-file individual made a difference to the evil that occurred”. Therefore, each “bomber can truly reply to the victims or their survivors, ‘Why blame me? I have not caused your suffering’”.⁶

The Dresden case is supposed to illustrate three quite different ways in which causal and moral responsibility may come apart: an agent may be morally co-responsible for an event to a high degree even if her causal contribution to that event is a) very small, b) imperceptible, or c) non-existent. Therefore, Kutz concludes that principles of complicitious accountability should not be derived from “some deep metaphysics of causal responsibility”.⁷ In cases of collectively produced wrongs, the relevant link is not causal, but “logical”, “semantical”, “teleological”, or “normative”. In the present paper, I defend the causal involvement condition for co-responsibility against the type of objections mentioned above. These are my main lines of argument:

First, overdetermination cases can be accommodated within a theory of co-responsibility without giving up the causality condition. Kutz and others oversimplify the relation between counterfactual dependence and causation, and they overlook the possibility that causal relations other than marginal contribution could be morally relevant.

Second, harmful effects are sometimes overdetermined by non-collective sets of acts. Over-farming, or the greenhouse effect, might be cases of that kind. In such cases, there need not be any formal organization, any unifying intentions, or any other non-causal criterion of membership available. If we give up the causal condition for co-responsibility it is difficult to see how one could delimit the morally relevant set of acts related to those harms. Since we sometimes find it fair to blame people for such harms, we have another reason to be sceptical about the idea that overdetermination defeats the causality condition.

Third, although problems about imperceptible effects or aggregation of very small effects are morally important, e.g. when we consider degrees of blameworthiness or epistemic limitations in reasoning about how to assign responsibility for specific harms, they are strictly speaking irrelevant to the issue of whether causal involvement is necessary for complicity.

Over-determination and related causal complications figure in various counterexamples against the causality requirement for co-responsibility. Such examples and thought experiments belong to the standard arsenal in this context. However, these examples do not show what they are supposed to show, and that they are not really counterexamples to the idea that if you are co-responsible for an event, then you have been causally involved in its production. I find this to be an important point, regardless of whether we believe that the causality requirement for co-responsibility is plausible or not.

It is beyond the scope of this paper to develop a positive theory of the relation between causation and co-responsibility. In the last section, I nevertheless present some reasons for thinking that the burden of proof rests on those who want to reject the causality condition altogether. The costs of rejecting the causality condition for complicity are high. Causation is an explicit and essential element in most doctrines of legal liability and it is central in common sense views of moral responsibility. Giving up this condition could have radical and unwanted consequences for legal security and predictability. However, it is not only for pragmatic reasons and because it is a default position that we should require stronger arguments (than conflicting intuitions about mismatches) before giving up the causality condition. An essential element in holding someone to account for an event is the assumption that her actions and intentions are part of the explanation of why that event occurred. If we give up that element, it is difficult to see which important function responsibility assignments could have, or why we should care when we are held responsible.

The Term Co-responsibility

Before entering into the discussion about whether co-responsibility presupposes causal involvement, it seems necessary to reveal some background assumptions (somewhat simplified though, since these assumptions are not the focus of the present paper).⁸

Like most philosophers debating individual responsibility, collective responsibility and co-responsibility, I assume that the term “responsibility” in these three phrases express the same concept. (There are a few opponents to this view. D.E. Cooper, for instance, suggested that in assigning collective responsibility, we employ a distinct notion of responsibility that differs from the ordinary one by being unconnected to blame and punishment.⁹)

The differences between these three ways of assigning responsibility concern, instead, the type of agent that is held responsible: an individual, a group as such, or an individual in her capacity of being a group member. Moreover, they concern the basis for the assignment of responsibility. Like Kutz, Bratman and others I believe that there is a substantial distinction to be made between collective actions and mere sets of individual actions. When we hold a group as such responsible for an event, we do that because they have performed a genuinely collective action, which led to that event. When we hold an individual co-responsible for an event, we do that because of the features that make her a member of a group held collectively responsible for that event.

As Kutz says, there is no collective responsibility without complicity. My own main reason for agreeing with this is that like John Stuart Mill, Alan Gibbard and others,¹⁰ I am talking about responsibility in a thick sense, in which moral responsibility is essentially connected to the justifiability of blame and other moral sanctions. Since collective sanctions strike individual members, our method for delimiting the collective agent that is morally responsible for a specific harm should be such that it picks individuals that justifiably can be blamed for what the group has done. In that sense, I think that fair assignments of collective responsibility should presuppose fair distribution of co-responsibility. We need a link between the individual and the group’s act, allowing us to separate co-responsible members of the relevant group from innocent bystanders. (This view has its opponents as well. Margaret Gilbert, for instance, thinks that judgments of collective responsibility have no implications either way for judgments about members’ responsibility.¹¹)

You are individually responsible (in the standard sense) for your intentional contribution to some harm—normally for the difference you intentionally made to the occurrence of that harm. But that sort of individual responsibility need not coincide with your co-responsibility for the overall harm produced by the collectively responsible group.

To hold you co-responsible is to hold you to account for the group’s act in virtue of the features that makes you a member of the collective agent. The direction of the relation between collective responsibility and co-responsibility is top-down—members will be morally tainted by the moral worth of the collective action. This may be contrasted with the direction of the relation between attributions of individual responsibility for intentional contributions, and a mere aggregation of those attributions.

The issue that I discuss in the following is whether individual causal involvement in the group’s production of some event for which we hold the group responsible is one of the features that are necessary to make someone co-responsible for that event. More specifically, I try to show that a certain type of common argument against the causal involvement condition for co-responsibility does not hold.

Overdetermination Does not Defeat the Causality Requirement

Kutz makes a distinction between principles that concern the basis of accountability (what makes an agent accountable for some harm), and principles about the object of that accountability (what harm the agent is accountable for). Much of his discussion is an attempt to refute the Individual Difference Principle (IDP): Basis — I am accountable for a harm only if what I have done made a difference to that harm’s occurrence. Object — I am accountable only for the difference my action alone makes to the resulting state of affairs.

From the refutation of IDP Kutz concludes that causal involvement is unnecessary when it comes to complicitious accountability. His positive thesis is that participatory intentions ground co-responsibility — the relevant link is “normative” or “semantical” rather than causal.

My focus in this paper is on the basis of co-responsibility. Under which conditions is it morally acceptable to regard an individual as co-responsible for some harm produced by a group? Although it is not an issue that I will pursue here, I am inclined to think that Kutz’s characterization of what he calls the “object” version of the Individual Difference Principle is unclear, and that one quite reasonable interpretation of his description will make the distinction between the “basis” and “object” versions of IDP collapse. I would prefer to distinguish between questions about the basis of accountability in Kutz’s sense (the conditions that must be fulfilled in order for an agent to be accountable for some harm) and questions about how and to which degree that agent should be blamed or punished on account of his fulfilling those conditions.

The strongest claim Kutz makes about the Dresden bombings is that no individual bomber made any difference. That might be correct. If the result was over-determined then each individual bomber could truly say that his specific bombings made no difference. The resulting event would have occurred even if he had abstained from bombing. However, from that it does not follow that he would continue to speak the truth if he went on to tell the victims “I have not caused your suffering”.

Although Kutz never explicitly subscribes to any specific theory about causation, his reasoning presupposes that if my act made no difference to the occurrence of an event, then I was not causally responsible for the event. Without that presupposed conditional, he could not conclude from the rejection of IDP that the relevant link “must be one of meaning rather than causality”.¹² That does not commit him to a full-blown counterfactual theory of causation, but at least to the view that counterfactual dependence is necessary for causation: ‘C caused E’ implies ‘If C had not occurred, E had not occurred’.

The individual bomber who believes that he caused no suffering simply assumes that if the event would have occurred in the absence of his action, then he cannot have been causally involved in its production. This way of thinking is not uncommon when applied to more straightforward causal processes. In this case, the trouble is that each bomber can truly say that the event would have occurred in the absence of his specific act. So, it seems that if causing an event requires making a difference to the occurrence of the event, no one caused the event in question.

In his impressive work Causation and Responsibility, An Essay in Law, Morals, and Metaphysics from 2009, Michael S. Moore joins earlier critics of David Lewis’s counterfactual theory of causation in dismissing the assumption that causation implies counterfactual dependence. Like the previous critics of Lewis, he rejects the counterfactual account precisely because of difficulties in handling over-determination cases.¹³ If that dismissal is part of the correct view on causation, the (im)plausibility of the Individual Difference Principle is irrelevant to the question of whether causal involvement is required for accountability. Unless counterfactual dependence is necessary for causation, the individual bomber could not truthfully deny causal involvement merely in virtue of the fact that the event would have occurred without him.

But even if we hold on to the counterfactual dependence requirement for causation, it can be refined to accommodate over-determination and similar difficulties.¹⁴ Kutz notes briefly that “we can stipulate solutions to the relevant metaphysical causal riddles”, although he thinks that these stipulations would not help illuminating the ethical questions.¹⁵ The solution he has in mind rests on the assumption that events are fragile, as Lewis puts it. “We may say that the actions of each bomber was necessary to that particular pattern of destruction”.¹⁶ Michael S Moore calls this “the fine-grained individuation of effects strategy.”¹⁷ If any bomber had abstained another pattern of destruction, and hence another event, would have resulted. So, the actual event ‘The Dresden Bombings’ would never have taken place if just one of the actual bombers had been missing (or some other minute detail had been different). (On this version of the “fine-grained strategy” there are not really any genuine overdetermination cases.) As Kutz says, “metaphysical differences are not (always) moral differences” and it would seem absurd to hold a single pilot fully responsible for the total event. Unlike Kutz, I nevertheless think that this way of reasoning could be relevant to the accountability project.

As Lewis stresses, there are no firm conceptual criteria in ordinary language when it comes to the identity of events. Especially when we talk about large-scale events, we can think of slightly different versions of them, with slightly differing chronological and causal properties.¹⁸ In most contexts it would be linguistically inappropriate to individuate ‘The Dresden Event’ in a way that includes only one version of it. You and I might disagree about the exact hours and details of the event, without doubting that we refer to the same thing.

As Lewis makes clear (developing an amendment of the counterfactual analysis, suggested by G.E. Paul), the counterfactual account can still accommodate cases like the one described by employing the notion of fragility. We should just “relocate the fragility not in the event itself but rather in a tailor-made proposition about that event, that will be a proposition about how and when and whether the effect occurs”.¹⁹ We should not understand ‘C caused E’ as implying that ‘if C had not occurred, E had not happened,’ but rather ‘if C had not occurred, E had not happened, or E had not happened in the same way or E had not happened at the same time’. The individual bomber’s causing the Dresden Event does then not only depend on whether this event would occur without his bombing, but also on whether it would occur at exactly the same time and in exactly the same manner without it. This second way of locating fragility admits of over-determination in a sense. It could be true that the same event would have occurred in the absence of one of the actions that actually caused it. It could just not be true that it would have occurred exactly then or exactly in that way.

By now, we have turned the term ‘caused’ into a rather technical term, where it appears anything that makes a difference to any detail of an event could cause that event. “You caused that!” would in daily life suggest that you alone caused it, or at least that your contribution was salient or significant, perhaps the “triggering” cause among a host of necessary causal conditions. Less misleading phrases for the causal relation allowed by the conditional dependence requirement would be “You were causally involved in it!” or “You were causally co-responsible for it!”

Even with that proviso, the criterion might seem to “open the gate to a flood of spurious causes”.²⁰ If some mother had not posted the letter that was in one pilot’s back pocket, the Dresden bombings would not have occurred in exactly the same manner, so it would seem that she could be causally co-responsible for the Dresden event. One of Moore’s main reasons for dismissing the fine-grained defence of the counterfactual condition is precisely that it “generates enormous promiscuity”.²¹

I believe this reason for dismissing the fine-grained strategy is mistaken. In the present discussion, just like in Moore’s own section on the fine-grained strategy, counterfactual dependence is merely supposed to be necessary for causal involvement, not sufficient.²² So, although the requirement taken on its own would be compatible with our finding the mother causally co-responsible, it would not at all force us to do so.²³ Nor would it force us to the conclusion that the degree of termite infestation (which affected the temperature of the fire) was a cause of the house-burning, on a par with the defendant’s negligent starting of a fire, as in Moore’s parallel thought experiment. (The fine-grained strategy would result in promiscuity on the assumption that counterfactual dependence is sufficient for causation, but that is beside the point here.)

Moore has two more complaints against the fine-grained strategy but they do not affect the point made here. First, he thinks that the strategy requires a Leibnizian notion of identity to be applied to events across possible worlds. This notion, he argues, is too strict when applied in everyday talk about identity. That, however, is an argument against the idea that events are fragile and it does not affect Lewis’s second version of the fine-grained strategy, which does not locate fragility in the events, but in a proposition about whether, when and how the event occurs. That second proposal admits of the possibility that a specific event could have occurred even if it had been slightly different. Second, Moore simply questions the assumption that an event could not have occurred in exactly the same manner, down to the most microscopic level, if it had been produced by another cause than the actual one. As far as I can see, this is just another way of expressing the negative view that our concept of causation involves no kind of claim about counterfactual dependence, rather than a new argument for that view. Anyway, if this negative view were the correct one, then the fact that a certain event would have occurred exactly in the same way even if I had not been there would certainly not let me conclude that I did not cause the event in question.

This leaves us with three main ways of conceiving the matter. First, Moore and others could be right about counterfactual dependence being unnecessary for causation. In that case, the mere fact that I made no difference as to whether, how or when an event occurred would not automatically exempt me from causal responsibility for that event.

The second possibility is the one Kutz hints at: If we regard the Dresden event as maximally fragile, then each bomber made a difference to its occurrence, and we could invoke the Individual Difference Principle to hold him co-responsible. As Kutz, Moore and Lewis all note, that is not how we would individuate this type of event in daily speech. Nor is it how we would understand “making a difference” in an ordinary moral context. However, this way of thinking would not necessarily have the absurd moral implications that Kutz thinks it would. Even if we applied this exacting interpretation of “making a difference” to the basis component in IDP, we would not be forced to reason in the same way when it comes to distributing blame among those causally involved in the harm in question. We could claim that an act could not make a person co-responsible for the Dresden even unless the act was such that had it not been performed, the event (counting all the most minute details) would not have occurred, while leaving open the question of how much of the blame the person (who fulfils that condition) deserves, if any. Our view concerning the person’s fair share of the blame could, for instance, hang on whether and to which extent other conditions for causal involvement were fulfilled (remember that counterfactual dependence is merely a necessary condition for causation in the present context), what other event we believe would have occurred instead, and what the intentions were. In other words, although this version of the fine-grained strategy may be odd in relation to ordinary language, it would not imply that a single pilot should be held to account for all the destruction in Dresden.

The third way of looking at the matter would be to hold on to the counterfactual condition but admit that even if the event had not been caused exactly in the way it was, it might still have occurred. So, we could say that the Dresden event would have occurred in the absence of one of the bombers, albeit not in exactly the same manner and at exactly the same time. Since the individual bomber makes no difference to the occurrence of the event in this case, we could not invoke the Individual Difference Principle in holding him to account morally. In this way, over-determination would defeat IDP. It would not thereby defeat the causal co-responsibility requirement, not even on the assumption that causal co-responsibility presupposes counterfactual dependence. The event would be robust, but we could still think of ‘how’ and ‘when’ as fragile. Each bomber makes a difference with respect to how and when the event occurs, so he could be causally responsible.

On either of these descriptions, no bomber can truly deny that he was causally involved in the production of the disaster. Although we may have to give up the Individual Difference Principle, we can therefore hold on to the view that causal involvement is a necessary component in the basis of co-responsibility. In the discussion of overdetermination cases, Kutz and others overlook the possibility that causal relations other than marginal contribution could be required in the basis of co-responsibility.

It is important to note the negative character of my argument so far. I am not claiming that the “fine-grained account” by itself captures the concept of causation, and I don’t think that the truth of “if C had not happened, E would not have happened, or E would not have happened exactly at that point in time, or E would not have happened in the same way, down to the most minute detail” is sufficient to make “C caused E” true.

My argument concerning over-determination cases proceeds from an assumption that is common to my opponents, who believe that over-determination cases defeat the causality requirement for co-responsibility. Their argument from over-determination presupposes that counterfactual dependence is a necessary condition for causation. One might reject that presupposition, as Moore does. In that case, over-determination cases will not be relevant to the issue of whether causation is necessary for co-responsibility. Or one might accept the presupposition. In that case one has to adopt a fine-grained approach to counterfactual dependence, or simply deny that there are any over-determination cases. Either way, over-determination cannot be used show what it is supposed to show in this context, namely that we have to abandon the causality requirement for co-responsibility.

Harms that are Overdetermined by Non-collective Sets of Acts

Kutz’s theory of collective action and intention fits into a tradition where thinkers like Michael Bratman, Margaret Gilbert, John Searle, and Raimo Tuomela, have drawn a substantial distinction between collective actions and mere sets of individual actions in terms of participants’ intentions. In Kutz’s version of this approach, “the core, minimalist, notion of collective action requires/…/that individuals act on overlapping participatory intentions”.²⁴ Implicit in the idea of a participatory intention is “a conception that one is doing one’s part in a collective project”.²⁵ Participatory intentions are a special class of intentions, “differentiated by their group-oriented content.”²⁶ The content of this type of intention is “irreducibly collective”.²⁷ This sort of analysis raises some methodological problems. Like Kutz, I nevertheless believe that this is one of the most promising types of approach to collective action so far.²⁸

The picture Kutz offers us when it comes to action-explanations in general appears to be broadly Davidsonian. Actions are causally explained by intentional states, as well as rationalized by the content of those states. Resulting events can be ascribed to an individual agent insofar as they are caused by her acts. In Kutz’s words, “Causal relations are presupposed in the paradigm case of individual, exclusive ascription.” ²⁹ The elevator’s arriving to me “can only be ascribed to me if my button push in fact caused the elevator to arrive”.³⁰

However, “within the modality of collective action, teleological relations play the primary role in relating actors to events”.³¹

We are properly held to account for the actions of groups (and of individual group members) in which we participate, because these actions represent our own conception of our agency and our projects.³²

So, causal relations are supposed to be necessary (but not sufficient) for individual accountability, while complicitious accountability is grounded purely in participatory intentions.

The theory says that individual ascription presupposes causal relations, that over-determination defeats such relations, and that complicitious accountability presupposes genuinely collective action — action explained by participatory intentions. Therefore, it will not cover cases where sets of non-collective actions over-determine harmful events. To take a standard philosophical thought experiment; A and B simultaneously shoot me, each shot is sufficient to kill me, and we stipulate that they have no participatory intentions. On the assumption that over-determination defeats causation and that individual responsibility requires causal involvement, no one can be held responsible.

It seems apparent to me that many morally relevant real life cases would fall in between the two criteria as well. It would, for instance, be farfetched to assume that the individual fisherman fishing for cod in Newfoundland was driven by an intention to participate in a collective project. It is also highly probable that the collapse of the cod stock there some years ago was over-determined. This economic and environmental harm would certainly have occurred even if one trawler had been missing. If accountability requires either participatory intentions or individual causal responsibility, no one would be morally accountable for such harms, on the assumption that over-determination defeats the causality requirement.

If you are not convinced by the Newfoundland example,³³ think of the large scale professional but illegal fishing, which alongside the legal fishing by EU now threatens to empty the waters outside West Africa.³⁴ Again, if over-determination defeats causation, causal involvement is a necessary condition for individual responsibility, and participatory intentions are required for collective responsibility, no one can be held responsible for the disaster that can be expected. Much important large-scale harm, like over-farming, or the greenhouse effect, appears to belong to this category.

These cases reveal an internal problem within Kutz’s theory. Regardless of whether my real life examples are convincing, surely we can imagine a case in which we want to hold people responsible for an effect that is over-determined by a set of non-collective individual acts. That is sufficient to show that Kutz would need to complement his theory with criteria covering such cases.

Less conclusively, I believe these cases point to a more general problem for any theory of responsibility incorporating the idea that over-determination defeats the causality requirement. Evidence of causal involvement seems indispensable when we want to delimit the relevant set of agents that could come into consideration for liability in the first place. We will have to start with the overall harm that has made us morally concerned, and then try to find out who were causally involved. There need not be any formal organization, any unifying intentions, or any other criterion of membership available. So, these cases are not just problematic within the specific framework suggested by Kutz.

The Irrelevance of Problems About very Small or Imperceptible Contributions

If my act made a small, negligible or imperceptible difference to the occurrence of a great harm, it made a difference to the occurrence of that harm. All such case-descriptions imply that the counterfactual dependence condition is fulfilled; If the case-descriptions are true, they pose no threat to the idea causing something requires making a difference to its occurrence, nor to the idea that being co-responsible for something requires making a difference to its occurrence. In other words, they are irrelevant both to the question of whether marginal contribution is necessary for co-responsibility, and to the issue of whether counterfactual dependence is necessary for causation. At most, they show that degree of co-responsibility for a specific event need not correspond to the size of the causal contribution. This paper’s section on small or imperceptible harms could end here.

However, it might seem odd to insist that causal involvement is an essential and important condition for co-responsibility, while admitting that there is not always any straightforward and detectible relation between causal contribution to an event and degree of co-responsibility for that event. Especially within the consequentialist tradition, there has been much discussion concerning imagined cases that exhibit structural features similar to those Kutz assigns to the Dresden bombings.³⁵ I have discussed this issue at somewhat greater length elsewhere³⁶ but I think that some reflections on these cases are in place in the present context.

Consider, first, problems about aggregation of very small causal contributions. According to Kutz, each bomber’s contribution in Dresden was “marginal to the point of insignificance”.³⁷ To begin with, our intuitions about small contributions are easily misled when we think of horrible or large-scale harms, like a torture-victim’s pain, or a firestorm. Assume that a person should be blamed in exact proportion to the size of her causal contribution to the harm for which she is blamed. What she should be blamed for would still vary not only with her share of the total event but also with the value of that event. 1/8000th of an atrocity could be an atrocity. A contribution being minor in the sense that it causes just a small fraction of the total harm does not mean that it must be minor in terms of the magnitude of the specific harm that it causes. In cases like these, there is a risk of conflating size of contribution in the absolute (and relevant) sense with size of contribution in terms of quota.

More importantly, the causal involvement condition does not imply proportionality between blame and causal contribution. Consider two opposing views about the nature and justification of assignments of responsibility. One familiar way of thinking, often associated with retributivism and/or incompatibilism, is to assume that when I hold you responsible for some harm, I make a statement of fact, perhaps of the following kind. “You intentionally realised that effect, knowing that it was harmful, although you could have done otherwise.” On such a view, when we express our moral disapprobation of your deeds, we are ascribing to you properties such as rationality and autonomy, besides stating that you actually realized the effect in question, and that this effect was bad. According to the opposite position, blame is not primarily about ascribing properties at all, but rather, as Hume, Strawson, Gibbard and others argue, an expression of a certain type of social emotion and/or desire. On that view, assignments of moral responsibility reflects a demand for a certain complex type of emotive, conative, and social response: basically a desire that the other feels guilt. When you feel guilty about some harm, very roughly put, you feel bad about it in a way dependent upon your demand for goodwill from others, and you wish you had not been involved in that harm.³⁸

Regardless of whether we adopt the descriptivist or the expressivist account, the justification of an assignment of blame and responsibility will have to appeal to a variety of factors in addition to the claim that something bad has happened and that the recipient was involved in it. Such a justification must be sensitive to the agent’s type of involvement, the agent’s mental capacities, beliefs and intentions, and, at least within the second type of approach, the social context. It seems reasonable to think that cases of participation in collective action typically differ from ordinary individual actions in several of these respects. Your participation in a collective project signals a certain kind of commitment, and it shows that you have social bonds to other people with similar intentions, etc. Such considerations may explain why we might consider you highly blameworthy even for a very small contribution to a collectively produced effect. In that sense, the relative weight of other non-causal considerations typically becomes greater in cases of participation than in cases of single individual actions.

Again, that sort of admission does not undermine the causal involvement condition for co-responsibility. We can accept Kutz’s claim that each Dresden bomber’s causal contribution is near or below the level of moral significance, and hence conclude that the aggregated values of all individual causal contributions cannot add up to the great negative worth of the aggregated effect. Still, what makes the bomber co-responsible for the event in the first place is that his act contributed to its occurrence.

In his arguments about the causal involvement condition Kutz stresses, also, that the difference made by each bomber was imperceptible,³⁹ and he notes the “difficulty of tracing causal connections between individuals and atrocity”.⁴⁰

The problem about imperceptible harms is illuminated by Derek Parfit’s well-known “Harmless Torturers” thought experiment.⁴¹ Parfit himself (like Glover, Tännsjö and others) discusses whether there can be harms that cannot be felt by the victim of the harm in question, like when the effect of each act from each individual torturer is imperceptible for the victim in Parfit’s well-known example. Parfit has taken two positions on this issue. The view he seems to favour in Reasons and Persons is that there are imperceptible harms in this sense. Each of the 1000 individual torturers’ acts does harm to the victim — the torture victim’s state gets imperceptibly worse for each act. That is what the example forces us to say, on the two assumptions that the victim’s condition in stage 1000 is far worse than his state in stage 1, and that “not worse than” is a transitive relation. In that case, the act is wrong due to its making a difference to the occurrence of physiological changes in the victim, changes that constitute harm that cannot be perceived by the victim. Personally, I find this view to be the correct one. I see no reason why we should not allow changes such as ‘being brought closer to the perceivable pain threshold’ to be labelled ‘harmful’ just because the victim cannot perceive them. But if harms can be imperceptible, then each bomber can make a direct difference to the amount of harm produced, even if that difference in the amount of harm is imperceptible. The Individual Difference Principle would not be affected.

The view on the Harmless Torturers example that Kutz assigns to Parfit⁴² is the one Parfit prefers in his ‘Comments’ (1986) – maybe harms by definition must be perceptible, as a hedonist might claim, but each torturer’s act is nevertheless wrong since it belongs to a set of actions which together cause harm. Suppose for the sake of argument that we reserve ‘harm’ for damages that can be felt by the victim, and that the isolated effect of each Dresden bomber’s act could not be felt by any victim (although the aggregated effect was felt by many victims.) This description is fully compatible with the assumption that though the isolated effect of each individual contribution was harmless in itself, it still constituted a marginal causal contribution to the aggregated effect that was extremely harmful. Again, this would not imply a mismatch between co-responsibility and causal contribution.

In the discussion following Reasons and Persons, “imperceptible” has sometimes been used in a quite different, epistemic, sense. A harmful effect of an act is imperceptible in this epistemic sense if it is impossible to trace the link between the act and the relevant harm. There is a “difficulty of tracing causal connections between individuals and atrocity” as Kutz says.

Writers like Shrader-Frechette (1987) and Gracely (1989) deny that, in principle, there are any imperceptible effects in this sense: “the effects of every non-mental act are capable of being known in some way, at least on the molecular level through sophisticated instrumentation”. If they were right, each bomber’s impact on the Dresden event would necessarily be perceptible, at least in principle. On the other hand, Sven-Ove Hansson and others have made a strong case for the existence of effects that are imperceptible in the sense used here. Hansson has also argued convincingly that such effects can make an act wrong. ⁴³

Regardless of whether it makes sense to speak of effects that are undetectable in principle, it seems evident that there are many real life cases in which the causal relations between specific actions and specific harmful effects are unknown to us, even when we have reasons to believe that there are such relations.⁴⁴ This may be more common when we consider individual contributions to collective projects but the problem can arise concerning individual actions as well. Such cases worry us, precisely because causal involvement is an essential element in our assignments of responsibility. Surely, the reasonable reaction to that lack of information is not to relax the causation requirements for co-responsibility, but to refrain from assigning moral responsibility or at least seek more evidence before doing so.

So, causal links between effects and act can be morally relevant, even in cases where it is impossible from a spectator’s point of view to expose those links. The existence of such cases does not show that mere imperceptibility of a causal link makes that link morally irrelevant. However, if we cannot expose the causal connections that make the act wrong, and we have no other evidence for assuming that the act had anything to do with the event that made us morally concerned in the first place, we are simply not in the epistemic position to pass moral judgment. That sort of restriction on our moral practices seems to be trivially required by any reasonable moral position.

Why Causal Involvement Matters

Examples concerning over-determination, very small effects, and imperceptible effects, figure in popular arguments against the causality requirement for co-responsibility. Such examples and thought experiments belong to the standard arsenal in this context. I have tried to show that these examples do not show what they are supposed to show, and that they are not really counterexamples to the idea that if you are co-responsible for an event, then you have been causally involved in its production. As I have claimed, I find this to be an important lesson, regardless of whether we believe that the causality requirement for co-responsibility is plausible or not.

It is beyond the scope of this paper to develop a full-blown positive theory of the relation between causation and responsibility. To begin with, that would require a positive analysis of concepts like ‘causation’ and ‘causally involved’. Note that the only commitment I have made concerning the concept of causation so far is conditional; Even if counterfactual dependence is a necessary condition for causation, we can accommodate the existence of causal relations in over-determination cases, since there is at least one plausible version of the fine-grained account of counterfactual dependence.

Let me, anyway, present some reasons for thinking that the burden of proof rests upon those who want to reject the causality condition for co-responsibility. In doing so, I have to employ an informal concept of causation. Any positive analysis of the concept of causation, whether it contains a counterfactual dependence condition or not, must makes sense of the explanatory role that is the primary function of the concept. When we seek for causes, we want to know why things happened. When we ask about your causal role for a certain event, we want to know whether your actions were among the elements in the process leading to that result, or part of the explanation of why that event occurred. To cause something is to make it happen, according to OECD. I assume that we can understand phrases like these without committing ourselves to a specific theory of causation.

If we give up the requirement of causal involvement for co-responsibility, the most obvious remaining criterion will be intent. As we have seen, Kutz explicitly advocates that single criterion when it comes to complicitious accountability.

Kutz stresses that complicitious accountability makes sense within “particular concrete social relationships”. He concludes that principles of complicitious accountability should not be derived from “some deep metaphysics of causal responsibility”.⁴⁵ “The salient connection between agent, act, and victim is essentially one of meaning rather than causality”.⁴⁶

This is a more radical position than it might seem at a first glance. As Margaret Gilbert and Tracy Isaacs in different ways have pointed out with reference to Kutz, it would seem absurd to blame or convict someone for an event that never took place, even if the accused had all the right sorts of participatory intentions.⁴⁷ Perhaps that specific criticism is too harsh. Maybe we should assume as an implicit background assumption in Kutz’s theory that assignments of co-responsibility for some harm must be restricted to harm that actually occurs. Even so, giving up the causality condition is a very counterintuitive step. As Michael S Moore forcefully shows in Causation and Responsibility, causation is an explicit and essential element in most doctrines of legal liability and it is central in common sense views of moral responsibility.⁴⁸

Moreover, it could be argued that our epistemic capacities are even more limited when it comes to determining intent, than they are when we try to determine causal connections. A legal system assigning co-responsibility merely on the basis of intent would hardly be secure or predictable.

Altough adequate causal involvement is a central requirement for guilt in modern legal systems, it must be admitted that there are exceptions to this rule. Most of those exceptions can be found in tort law, where the need to demonstrate causal responsibility can be abandoned in some types of cases. However, the focus of tort law is on compensation and prevention, rather than on punishment and desert. In tort law it is sometimes necessary to lay aside questions of fault and culpability for the benefit of long-term safety, or just to guarantee that costs for compensation are covered in a reasonable way. It is noteworthy that most exceptions to the requirement of intent also can be found in tort law, where it is considerably more common to apply the notion of strict liability than in criminal law. So, we should not expect tort law to be completely in line with people’s moral intutions about guilt and responsibility.

There appears to be a few exceptions in criminal law as well, especially regarding the types of acts that are in focus in this paper.⁴⁹ In some countries, a person can be convicted of murder because of participation in a joint act, even if it is clear that other persons have caused the actual death of the victim. I would hesitate to draw any conclusions about common sense views from this fact. To begin with, legal practices concerning collective actions differ substantially between countries, even when those countries share most other central legal and moral features.⁵⁰ Moreover, cases like these usually stir up controversy when they become publicly known – some protest when mere participation is treated as being on a par with causal involvement, while others are just as upset when a group member is let off the hook just because he was not causally involved. It is natural to think that such controversies depend precisely on those issues that are under debate here – the difficulties of sorting out causal relations, and the opposing moral intuitions we have about group actions.

Of course, the mere fact that the causation condition is default in legal and moral practices is not a substantial philosophical argument for holding on to the condition. Nor are mere pragmatic considerations about legal security. But at least they indicate that the burden of proof rests heavily on those who want to reject the condition. Stronger arguments than conflicting intuitions about mismatches between causal and moral responsibility in causally complicated cases appear to be needed.

In a defence of Kutz’s rejection of the causation condition, Brian Lawson suggests ways of meeting standard intuitions about causal involvement and responsibility.⁵¹

First, he points out that we do not always assume that the degree of an agent’s responsibility for an event is proportionate to her causal contribution to that event. I agree, but as I claimed in the section on very small causal contributions, assignments of “degree of responsibility” are most plausibly understood in terms of degree of blameworthiness. The question of how strongly we should blame a person for a specific harm depends on many other factors, besides causal involvement. So, this disproportionality is not really relevant to the question of whether causal involvement in an event is a necessary condition for being co-responsible in the first place.

Second, he notes that in cases where there is intent but the intended action never takes place (say, because the person intending to murder hits her head and forgets about the whole project), we may have no evidence of intent. “[E]vidence of success, or at least evidence of near success, affects the nature of our intuitive responses.”⁵² So, our intuitions about the importance of causal involvement could stem from our regarding effects as indicators of what is morally relevant, namely intentions. Although Lawson does not make this point, one might add that the indicator function might justify keeping the causal requirement in legal contexts, for mere purposes of legal security.

To begin with, this move brings back all the worries that made Kutz and Lawson reject the causality condition to begin with. If problems with negligible contributions, imperceptible causal links or overdetermination should make us sceptical about the causation condition, they should make us just as sceptical about the possibility of using effects of intentions as evidence of intentions, at least when we consider collective actions.

As we have seen, Kutz requires causal involvement for responsibility in the individual case but not in the collective one. Similarly, Lawson believes that our intuitions differ between the collective and the individual case, when it comes to requiring causal success. “It seems a plausible intuitive fit to say that once someone commits to contributing to an outcome along with others, her being blameworthy for her participation is not as sensitive to the extent of her causal success as it would be in a straightforward individual case.”⁵³ I don’t share that intuition. If we have evidence of an agent being committed to contributing to an outcome along with others, but it is clear to us that the agent completely fails to contribute to that outcome, I would regard it as absurd to blame the agent for that outcome. The case would simply be parallel to Lawson’s interrupted murderer case, where we know that the agent is committed to contributing to an outcome on his own, but fails completely.

In a television comedy sketch I watched recently, the thoughtful mother of a blind young neo-Nazi regularly swaps the son’s swastika-badges and Nazi-decorations for completely innocent symbols, without his knowledge. Suppose this son joins a neo-Nazi-demonstration and that the demonstration to some minor extent is successful in creating conflict and violence. At the same time, unintentionally the blind son radically diminishes this harmful effect of the collective behaviour, just by being visible in the crowd with cute symbols on his clothes. (Perhaps it is difficult to imagine that the boy’s presence in the demonstration will not contribute in any way to the harm caused by the group. Insofar as that difficulty makes one more prone to hold the boy responsible for that harm, it reveals that one has intuitions about the importance of causal involvement.) He fully shares the participatory intentions of his fellows but he obstructs the fulfillment of those intentions. We should blame him for trying but surely we cannot blame him for the small raise in conflict and violence that is an effect of the demonstration. It turned out that his mother’s well-intentioned deception successfully prevented him from making himself responsible for that sort of harm.

The idea that causal links are relevant merely as indicators of intentions gets things the wrong way around in an illuminating way. Causal involvement is essential to the role assignments of responsibility and co-responsibility play in our moral practices. When we blame people and hold them to account for some harm, we involve them in our moral sphere. That sort of involvement is meant to make them react with corresponding feelings of guilt or remorse, not over past states of mind as such, but over what these states of mind have led to. Resulting feelings of guilt, unlike mere shame, motivate “by sticking to things where motivation could make a difference”, as Gibbard says.⁵⁴ We want to make them realize that their choices had an impact – that it was no coincidence that bad things happened when they made those choices. Their choices mattered, not by themselves, but because of their connections to events that made the world worse. If moral condemnation did not come with the implicit assumption that the harm for which you are blamed in some way was explained by your choices, why would blame make you reconsider those choices at all?

Summary

When we are confronted with certain types of causal complications in situations that call for our moral judgment, it is tempting to relax or give up the causal involvement requirement for co-responsibility. In one sort of cases, we feel that some or someone certainly must be responsible for some harm, although no individual action made a difference to the occurrence of that harm. Other counterexamples to the causal involvement requirement concern discrepancies between size of individual contributions and the share of the blame we feel that each contributor deserves. In a third type of cases the difficulties are epistemic: the relevant causal links between act and effect are undetectable or the specific effects of each individual act are imperceptible.

If we give in to that temptation and abandon the causal involvement condition for responsibility for such reasons, it will be difficult to handle causally complicated cases where the contributors to some harm have nothing in common apart from their causal involvement in that harm. Much large-scale harm appears to fall into this category.

However, the various standard counterexamples do not prove that causal involvement is unnecessary for co-responsibility. Over-determination cases show that we should adopt a more subtle view of the relation between counterfactual dependence and causation. Disproportionality between size of causal contribution and intuitions about fair share of blame show that justifications of blame should be sensitive to a variety of factors besides causal involvement – most notably, intent. Imperceptibility of specific effects, or undetectability of causal links between acts and effects, should make us cautious before assigning responsibility, precisely because the causal involvement condition is an essential and practically important element in our practices of holding agents morally to account for events that has made the world worse. All these admissions are compatible with the assumption that if you are justifiably held co-responsible for some harm, then you have been causally involved in the production of that harm to begin with.