Complicity and the responsibility dilemma

Abstract

Jeff McMahan famously defends a moral inequality of combatants, where liability to be attacked and potentially killed in war, should be grounded in the individual combatant’s moral responsibility for posing an unjust threat. In a response, Seth Lazar shows that McMahan’s criterion for liability leads to an unacceptable dilemma between “contingent pacifism” and “total war”, i.e. between war being practically infeasible, or implausibly many civilians being legitimate targets. The problem is that McMahan grounds liability mainly in the individual’s causal responsibility for posing an unjust threat, but where a large proportion of combatants and civilians are approximately equally causally responsible. Recently, Saba Bazargan has come to the aid of McMahan by injecting an alternative supplementary criterion for liability, namely the individual’s complicity in a group act. This criterion is supposed to uphold the noted moral inequality, while avoiding the responsibility dilemma, by grounding moral incrimination in the individual’s participatory intention, instead of her causal contribution. I argue that the complicity account fails to resolve the dilemma. It fails because complicity grounded in a causally inert participatory intention is insufficient for liability. Further, I show why this reveals a deeper problem with the complicity account itself, namely that though it purports to ground incrimination non-causally, it fails to do so to any serious extent.

1 Introduction

A long standing assumption in the just war tradition has been that there exists a moral equality of combatants.¹ Specifically, combatants on both sides of a war have a symmetrical right to intentionally attack, and kill one another, regardless of the justification for the war itself. Recently this assumption has been challenged and sought revised. Most prominently, Jeff McMahan argues that the moral principles governing the practice of war ought to be an extension of the norms governing society at large (McMahan 2009: 14). In society at large, we accept that everyone has a default, almost absolute right, against being intentionally targeted, unless they have alleviated themselves of that right. One way of doing so is by being morally responsible (MR) for violating other people’s right by harming them or posing a threat of wrongful harm to them. The particular account of MR McMahan subscribes to is in line with standard control based accounts of MR.² For an agent to be MR on his account, she has to have performed a voluntarily³ action, which resulted in her being causally responsibility for the harm or threat, where the harm or threat was minimally foreseeable.⁴ In war, this then leads to a scenario where a soldier of side A may potentially be justified in attacking and killing a soldier of side B, but not vice versa. E.g. if the soldier of B is MR for posing an unjust threat to the soldier of A (or her side), by voluntarily making a somewhat foreseeable causal contribution of harm or threat of harm to A (or her side), B has relieved herself of her right against deliberate attack, and B may then be attacked and potentially killed in order to mitigate B’s personal threat, or B’s side’s overall threat.

Though many agree with McMahan’s broad revisionist conclusions, a problem arises when we apply McMahan’s particular MR criterion for liability. As Seth Lazar has argued, it leads to a responsibility dilemma where we have to choose between two equally unacceptable scenarios, namely “contingent pacifism”, or “total war” (Lazar 2010: 189). The problem is that if we accept McMahan’s criterion, a large proportion of civilians and combatants will fulfill the conditions for MR to the same extent. The issue lies with the fact that many civilians are causally responsible for posing at least as large (or as small) a threat as many combatants, simply in light of going about their lives as members of society e.g. by paying taxes, voting, and raising their children in a patriotic or religious atmosphere (Lazar 2010: 192). Even when their actions fail to constitute a personal unjust threat, they are minimally aware that they are enabling that threat, and they are thus responsible for contributing to their sides’ unjust overall threat. On the other hand, many combatants will remain far detached from combat, fulfilling support roles (e.g. logistics and communication), which are aggregately important, but individually causally insignificant. Or, they are simply causally ineffective at doing their job in part due to lack of skill, or an unwillingness to harm enemy combatants.⁵ Thus, a large proportion of combatants, and a large proportion of civilians, are at least approximately equally causally responsible for posing an unjust threat in war. Assuming they fulfill the rest of the conditions for MR to the same extent,⁶ this either leads to a situation where too many are MR, or too few, depending of the stringency of the causal responsibility condition. If it is too stringent, an insufficient number of combatants will be MR for war to be practically feasible, because we would be unable to attack anyone without also producing a disproportionate degree of collateral damage to non-liable individuals. If it is too yielding, we have to accept that many civilians are just as liable to be killed on the unjust side in light of their contribution, a conclusion which is intuitively unacceptable to almost everyone. Therefore, assuming Lazar’s critique holds, individual MR for posing an unjust threat is not an acceptable criterion for liability, because it fails to constitute an acceptable practical demarcation between civilians and combatants.

If we assume Lazar is correct, a possible way to salvage the moral inequality of combatants—assuming it is independently plausible—is to respond by injecting an alternative or supplementary liability criterion. Such a response can employ the following strategy: (1): accept the uncontroversial claim that MR is in principle adequate grounds for liability. (2): Specify the causal responsibility condition for MR to be sufficiently stringent to exclude most civilians (while inadvertently excluding most combatants). (3): introduce a supplementary criterion which incriminates the remaining relevant combatants, but excludes most intuitively non-liable civilians. In line with this strategy, Saba Bazargan (2013) employs the ‘complicity account’ as the noted supplementary criterion in an attempt to catch the intuitively unjust combatants, but only the unjust combatants. On the complicity account he adheres to, it is the individual’s mere non-causal participation in an immoral group project that grounds her incrimination in that project. Her incrimination then makes her liable to be killed when doing so sufficiently alleviates the threat brought about by her and the associated participants. However, civilians on the unjust side are on his account not participants in an unjust war, and thus not complicit, and thus not liable.⁷ Therefore the responsibility dilemma is seemingly resolved.

Though seemingly promising, Bazargan’s solution fails. The reason it fails is due to a problem with the complicity account and how it purports to ground incrimination non-causally. As I show, the complicity account is only plausibly sufficiently incriminatory when it implies that the agent in question is causally responsible. This drastically limits the account’s applicability as a viable substitute for traditional accounts of moral responsibility. Further, with regards to the war scenario, the complicity account’s reliance on causal responsibility ends up reintroducing the responsibility dilemma.

2 The complicity account

The ‘complicity account’ was originally developed by Christopher Kutz (2000a, b). His account is situated within an increasingly expanding area of moral theory, which has emerged as a response to the problems of ascribing MR to individual agents in cases of collective involvement. These problems arise because traditional individualist accounts of retrospective MR are predicated on the individual’s distinct sphere of causal control, i.e. the individual’s ability to contribute to, or hinder a given harm.⁸ Classic problem cases e.g. include large scale harms such as oppression and genocide, where a large proportion of the harm in question seems unaccounted for by causally underpinned notions of MR [e.g. Isaacs (2006) and Lepora and Goodin (2011, 2013)]. But, the issue can also crop up in cases of small scale harms such as those involving non-instrumental apparent accomplices. Thus, relying solely on traditional control based accounts of MR, will seemingly leave us with an unexplained normative residue which seems to prompt an alternative response.⁹ Broadly speaking, aside from denying the normative weight of the perceived residue [e.g. Narveson (2002) and Szigeti (2014)], responses have taken either collectivist or individualist form. Collectivists tend to respond by showing how we can attribute MR to groups as groups [e.g. French (1979) and Pettit (2007)]. Individualists on the other hand, of which Kutz has been one of the most influential with regards to this issue,¹⁰ attempt to ground MR for group outcomes in the individual. Kutz’ particular approach is to ground MR in the individual’s mere non-causal participation in group harm, rather than her individual direct causal responsibility for it.

Kutz’ ‘complicity account’ is by his own words a minimal account of joint action, where the individual’s mere participation in the joint action incriminates her in the harmful outcome that action brings about. It is minimal, because unlike ordinary accounts of joint action, which tend to concern small scale group activities (such as dancing, singing a duet, going for a walk or painting a house), it does not require strong reciprocal awareness and coordination between the participants. For that reason, it less demanding, and more readily applicable to complex large scale cases featuring individuals who lack intimate knowledge regarding the other participants.¹¹ Specifically: “the core, minimalist, notion of collective action as such requires only that individuals act on overlapping participatory intentions” (Kutz 2000a: 4, emphasis his). The method for inclusion in the joint action, and incrimination in the relevant harm that results from that action, is the individual’s inclusive authorship. She has inclusive authorship in a harm when she acts on a ‘participatory intention’ directed at the harm, to the extent that her participatory intention overlaps with the participatory intentions of the other participants. When these intentions overlap, “the will of each [is] manifested in the acts of all” (Kutz 2000b: 141). The harm itself is the causal result of some subset of the participating individuals, but not necessarily of all of them (Kutz 2000b: 81).¹² The upshot is, as noted, that my incrimination in harm, does not hinge on my personal causal contribution to that harm. It hinges on my will being connected to the wills of more causally efficacious individuals, in the right way.¹³,¹⁴ Thus, this account potentially allows him to solve some of the noted problem cases.

The individually necessary and jointly sufficient conditions for incrimination on Kutz account are: they have to be minimally aware of their participation (Kutz 2000a: 27).¹⁵ Further, though he never touches on it explicitly, they presumably have to be of sound mind, as well as have acted voluntarily.¹⁶ Lastly, and most importantly, instead of being causally responsible for the outcome, the agent as noted has to act on a participatory intention towards the joint project. In Kutz’ original work, acting on a participatory intention is specified as acting on: “an intention to do my part of a collective act, where my part is defined as the task I ought to perform if we are to be successful in realizing a shared goal” (Kutz 2000b: 81). Further: “A participatory intention has two representational components, or sets of conditions of satisfaction: individual role and collective end” (Kutz 2000a: 10). When a multitude of individuals have the same shared goal, and they individually act on the intention to perform an action directed at that goal, they are then complicit in the particular joint action, and the harmful outcome it produces. Lastly, an action can be said to be directed at the shared goal when its function is to “foster [the] collective end” (ibid), regardless of whether it actually ends up being causally relevant to the success of that project. Saba Bazargan understands “acting with participatory intention” roughly along the same lines, though specifying that the individual has to act in accordance with a sufficiently central role in the joint project, rather than act with the intention of fostering the collective end (Bazargan 2013: 188). This allows Bazargan to distinguish between degrees of incrimination, where the more central the function of one’s role in a project, the greater degree of incrimination and potential liability in the project.¹⁷

It is now clear how Bazargan employs the complicity account to resolve the responsibility dilemma. He argues that the individual combatant is complicit in the unjust war, not because she is MR in light of having made a direct causal contribution to it as McMahan suggests, but rather because she acts on a participatory intention to play a role in the unjust war, which overlaps with the same participatory intentions of other participants, where a subset of all the participants produce the relevant harm (the unjust threat). Because her will is tied to the wills of the other participants, she is complicit in this harm and thus she is liable to be killed. But, since only combatants act on the noted participatory intention, only combatants are complicit. Further, just as with McMahan’s MR account, only combatants on the unjust side are incriminated, since only they engage in joint projects which produce unjust harm. Lastly, since we assume the causal responsibility condition for MR is sufficiently stringent to leave out most civilians, the responsibility dilemma is seemingly solved.

3 Cases

In this section I present two cases which are supposed to provide intuitive support for the idea that we can incriminate ourselves in a significant sense by way of our participatory intention, even when we fail to causally contribute to the harm in question. The first case is Bazargan’s, the second one is Kutz’. I will present them, and then in the following section discuss the complicity account in relation to the cases.

Heist: “a criminal mastermind puts together a plan for robbing a bank. She hires five individuals, each of whom agrees to participate in the robbery. The recruits are made aware that part of the plan is to kill the witnesses in the bank. The mastermind does not physically participate in the robbery - instead, she provides the plan, the layout of the bank, the equipment, etc. One of the recruits, J, is stationed on a second floor balcony above the bank, as a look-out. Her role is the least important. The mastermind would have commenced with the plan even without a look-out. Suppose that J is not a very effective look-out—in fact, she falls asleep on the job. Fortunately for the robbers, J’s incompetence has no negative effect on the robbery, though her participation does not causally contribute to the robbery or murders either. The plan succeeds, and two witnesses are killed […] though J causally contributed nothing, she bears some liability for the murder of the witnesses and the theft of the money.” (Bazargan 2013: 182–3)

Bazargan wants to show that we indeed have the intuition that J is morally incriminated in the project of robbing the bank, and the shooting of the witnesses, even though J lacks causal responsibility for the harm. Bazargan tests the strength of this intuition by asking whether it would be permissible to shoot and potentially kill J, if her death would result in the innocent lives of the bank-heist witnesses, being saved. Bazargan is testing whether J’s incrimination, in light of her causally impotent participatory intention to play a role with a central function in the heist, can justify J’s liability. I.e. we are testing whether she is liable to be killed because she poses a threat in conjunction with the other group members. If we assume killing her is the only way to spare the lives of the innocent hostages, we would presumably find this acceptable. Thus we have reason to accept that having a causally inert participatory intention to fulfill a central role in the group harm, is sufficient for severe moral incrimination. Further, to the extent that this scenario is relevantly comparable with the just war scenario, we also have reason to accept that the combatant can be liable to defensive killing, if she has the participatory intention of fulfilling a relevant role in the unjust harm. Indeed, since it is J’s participatory intention which incriminates her in the group project and threat, and not her individual causal responsibility for contributing to that threat, the complicity account is seemingly a plausible explanation to our intuitions.

I want to present another relevantly different case. It is a simple firing squad case, which is also supposed to show that the intuition that intention alone is incriminatory generalizes.

Firing Squad: Five gang members have volunteered to shoot an innocent prisoner. They are all aware of her innocence. They shoot, however, due to an inventory mix-up, one of the cartridges is a practice round, and never produces a bullet. The victim dies from the gunshots.

This example is appropriated from Kutz (2000b: 2), cut from a standard generic template common in ethics. For brevity’s sake, it is severely under-described, but sufficiently clear to provide relevant intuitions. Glossing over many details which will be discussed later, we presumably do have the intuition that all shooters are incriminated in the death of the innocent victim. Further, if a sniper has been placed with a clear sight to the firing range, we would presumably consider it permissible for her to shoot and kill any one of the participants, if this would stop the immoral killing of the prisoner. Further, if the sniper had only a clear line of sight to the inefficacious shooter, we would presumably also consider it morally permissible to shoot her, if it was a reasonable assumption that this would stop the shooting (e.g. by startling them). This again lends credence to the notion that an individual can be incriminated in a group act of posing a threat, and potentially be liable to be killed in light of that threat, even if that individual fails to make a causal contribution to the group act. This then lends credence to the notion that the shooter’s participatory intention to fulfill a role, is sufficiently incriminatory to ground liability, since the shooter fails to contribute causally to the killing of the innocent victim.

4 Discussing the complicity account

After having provided intuitive support for the claim that a causally inert participatory intention can ground severe incrimination and liability for a group harm, I will now argue that despite first appearances, we lack clear grounds for accepting this claim. As I show, acting on a participatory intention is plausibly only incriminatory when it implies either causal responsibility for harm, or causal responsibility for the threat/risk of harm. Thus, it is not clearly applicable to the problem-cases where individuals participating in group harm fail to be causally responsible, or as tie-breaker in cases where there is an equality of causal responsibility, like the just war scenario.

Recall, the complicity account requires that the incriminated agent acts with a participatory intention directed at fostering the collective end (Kutz), or acts in accordance with a role which is designed to have a contributory function (Bazargan). It is not sufficiently incriminatory to merely have a participatory intention the passive sense of implying a mere want, desire, aim, or non-committed/non-initiated plan. The individual plausibly, and as a matter of assumption for both Kutz and Bazargan, actually has to act on the intention. This ensures that the individual is actually a committed participant, rather than a mere sympathetic bystander. Unfortunately, the reliance on this intentional action causes problems for the complicity account. The problem is that requiring an action, reintroduces causal responsibility as a necessary condition for complicity. Indeed, this implication of causal responsibility hinders the complicity account’s ability to solve the problem cases, and hinders its applicability as a solution to the responsibility dilemma.

Bazargan never explicitly discusses what these actions are, aside from noting that they should be in accordance with fulfilling a particular role in the group project. This however leaves an open question regarding which actions fulfill which roles, and why doing so is morally incriminatory. Kutz on the other hand does at least briefly specify what those accompanying actions should be. Specifically, for Kutz, an action is performed in accordance with the participatory intention when it is either “instrumentally”, “normatively”, or “expressively” (Kutz 2000a: 11) related to the joint project. It is ‘instrumental’ when it results in a causal contribution to the project. Though this is a plausible method of incrimination which closely mirrors the causal responsibility requirement in traditional accounts of MR, it is however uninteresting for this discussion. The problem cases, as noted, are precisely those cases where the individual is unable to make a tangible causal contribution. Further, this condition will also dilemmatically be met by combatants and civilians to largely the same extent. With regards to acting with a participatory intention which is ‘normatively’ linked to the collective project, Kutz writes: “the relation might be normative if one performs one’s part because of norms internal to some group or institution that demand certain behavior (I wear a dark suit as an IBM employee)” (Kutz 2000b: 82). He elaborates further: “Merely wearing appropriate clothing is not what constitutes my willing participation in [some organization’s corporate culture], but rather wearing [corporation-specific clothing] with the intention of being part of that culture” (ibid). Though Kutz’ description does not specify that this action explicitly relies on the individual being causally related to the project, it is however unclear how it is supposed to be incriminatory without implying causal responsibility. If someone is working at IBM, or perhaps more relevantly, is a civil servant working in the German bureaucracy during the second world war, there is a sense in which the bureaucrat can be incriminated in the holocaust for intentionally doing her part in the overall maintenance of the bureaucracy, knowing that this ensured the effective capturing, transportation and ultimate murder of many innocent people. However, incrimination to any serious extent presumably presupposes she was making a causal contribution to the harm, or the threat of harm. Or, it presupposes that she was able to mitigate the harm, but deliberately failed to do so, i.e. a causally underpinned omission.¹⁸ If the bureaucrat did not contribute to the harm, and foreseeably could not do anything to counteract the harm, she is not clearly incriminated, even if she behaved generally in line with the German bureaucratic culture. For all incriminatory intents and purposes, she is merely a bystander, who happens to work in an office associated with the German bureaucracy. The bureaucrat might be deeply sympathetic towards the commencement of the holocaust, but it is difficult to see how her sympathies add anything with regards to incrimination, when they do not result in a causal contribution, or in causally bringing about a threat.

Of course, if the bureaucrat aside from working in the central bureaucracy also had a strong intention (e.g. desire, want) to contribute to the holocaust, despite being unable to do so, there is possibly a sense of readiness to participate, should the appropriate conducing circumstances obtain. So, is this readiness incriminatory? Presumably yes, but presumably only to the extent that this readiness is predictive of the individual ending up contributing to the harm, or some other harm in the future. But, then what grounds her incrimination is again her causal responsibility for posing a threat. It is not her acting with a participatory intention as when she conforms with the norms of the group when wearing a certain uniform at all. It is her predictive causal responsibility, a predictability based on her desire to contribute to a great harm and the probability that the relevant conducive circumstances will obtain.¹⁹ A participatory intention may be indicative of risk, but it is not necessarily indicative. One can act in accordance with the noted norms, and still be unable, or unwilling to pose a threat should the circumstances obtain. Unless of course those norms specifically pertain to making harmful contributions. However, in that case, it is still the actual or expected causal contribution which is incriminatory, not strictly that one is acting in accordance with a company norm.

Regarding acting with a participatory intention which is ‘expressively’ directed at a group venture, Kutz writes that a relation is expressive: “if by doing one’s part, one thereby exemplifies one’s membership in a group or participation in an activity, as when by voting I express my membership in a political community” (2000a: 11). Presumably what he means by ‘expressive’ is what is at the heart of the expressive theory of voting. This theory is a response to the problem of explaining why people actually vote in large scale elections, when the utility from doing so is insignificant. It assumes that when people vote, most of the utility they derive from that action, comes from what voting expresses, rather than from the instrumental (causal) effect their vote has on the electoral result.²⁰ If this is the notion of expressive participatory intention Kutz has in mind, then it is simply not incriminatory. The voter may be morally tainted in some way by expressing her opinion. However, if we really do accept that she has not also contributed to the outcome, it is not clear how she is incriminated in the outcome of the vote any more than the sports fan is “incriminated” in her team winning the match (discounting the minimal affect her action has on team morale).²¹ Holding a mild reactive attitude may be appropriate, but it would be absurd if expressing oneself, or performing non-efficacious actions with expressive content, can ground severe punishment and liability. E.g. yelling: “kill all the Jews” at a rally is certainly expressive of a reprehensible attitude towards the target of that chant, but it is difficult to see how it can ground any serious incrimination in the death of the German Jewry, unless it is also backed up some kind of actual or predictive contribution. Of course, some expressive actions are expressive in part due to their expected causal content. Killing someone may e.g. be expressive of my disdain for, and willingness to kill a given person. However, it is not strictly the expressiveness of my action which grounds my incrimination, but rather the causal contribution to the relevant harm. Expressiveness can also here be predictive of future behaviour.

Summing up this section: the complicity account fails to be clearly incriminatory, or it reintroduces causal responsibility as a necessary condition for incrimination as causal responsibility for the harm either directly, as an omission, or predictively (risk/threat). In the next section I will review the cases again, in order to ascertain why our intuitions supported incrimination, when the relevant agent seemingly was causally inefficacious with regards to the harm in question.

5 The cases revisited

In the previous section, I argued that we lack clear grounds for accepting that a non-causal participatory intention is sufficiently incriminatory. However, this contradicts the intuitions evoked in the previously presented cases. When I presented the two cases, I assumed the reader’s intuitions supported the verdict that the relevant agent was complicit in the respective ventures, i.e. in posing a threat to the witnesses in the heist, and in the killing of the innocent prisoner in the firing squad. Further, I suspected that intuitions also supported the permissibility of killing the incriminated agent, if this would save the innocents involved. This provided intuitive support for the notion that participatory intention alone could be incriminatory, since the cases were setup in a way, where the only seemingly incriminatory feature was the intention to participate. Thus, there is now an onus on me to demonstrate that contrary to our intuitions, those cases are either not cases of incrimination at all for the agent in question, or to the extent that they are, they are so because they imply causal responsibility. Further, since these cases are selected as prime examples of the successfulness of the non-causal complicity account, showing that they in fact plausibly rest on implicit assumptions of causal responsibility, will lend credence to the claim that complicity cannot be separated from causal responsibility.

I will now show that it is not strictly the participatory intention as a non-causal condition that grounds incrimination in these cases. Rather, it is plausibly an embedded notion of causal responsibility. Before proceeding, I need to note that it is not a problem for the complicity account that it implies either actual, or predictive causal responsibility. I concede that these are plausible modes of incrimination. However, causal responsibility will constitute a limitation to the complicity account’s applicability. Further, if the complicity account implies causal responsibility, it is not clear how it is supposed to solve the responsibility dilemma in the just war debate, since equality of causal responsibility was the main problem here.

Starting with the heist case, if we can explicate certain plausible details and clearly remove any trace of actual, omitted or expected causal contribution to the heist, our intuitions of J’s liability dwindle fast. This supports the claim that causal responsibility was indeed the main root of incrimination. Regarding J’s role, she is in a narrow sense clearly ineffective at her job, as stipulated. But, we would be hard pressed to assume that the other participants (and J herself) had good reason to believe so from the way the case is written. I.e. the other participants and the boss were presumably increasingly inclined to perform their roles, knowing there was a lookout in place, which would warn them and increase the probability of getting out unscathed. Even if their participation did not depend on J in the binary sense, her role is actually causally relevant in bringing about the heist since it actually contributes to sustaining the group itself and thus ensuring the effective commencement and success of the project. She is causally relevant to the project. E.g. on a simple counterfactual model of causation, it is true that they either would not have participated, or that they at least would not have had been as motivated without a lookout. J being a lookout is a relevant putative fact with regards to which they plausibly act and which optimizes the effectiveness of the participants. Therefore, their participation and their inclination is causally affected by J’s participation. Thus, J does make a causal contribution. In addition, J was plausibly aware, or should have been aware of how her role affected the cohesion of the group and the effectiveness of the other participants, and that her participation was causally responsible for increasing the likelihood of the witnesses getting killed, i.e. for posing a threat.²²

Further, despite what we are told, it is difficult to imagine the mastermind would have given the go-ahead for the mission without a lookout, or again, that the mastermind would have at least been as strongly inclined to go through the mission without one. If having a lookout does not affect her decision to go forward with the heist, why would she include a lookout to being with?! The fact that she actually did include a lookout pushes us towards the implicit assumption that having a lookout is preferable to not having one, i.e. J is causally responsible for part of the mastermind’s reasons for moving forward with the plan. Put in terms of probabilistic causation, having a lookout increases the likelihood of her going through with the mission. So, it is not simply J’s intention (aim) to play a role which incriminates her, rather, it is more importantly her foreseeing or predicting that her inclusion in the project is causally relevant. This is presumably at the heart of our intuitions in the case.

In order to test whether our intuition of incrimination is indeed driven by her causal contribution rather than an inert intention to participate, we can revise the case to more clearly stipulate her role as causally superfluous. While doing this, we update her own, and the other participants’ knowledge of her causal impotence. When we do this, and if our intuitions no longer suggest her moral incrimination, that suggests that the causal role was a necessary condition for her incrimination. So we keep the original case in mind, but explicitly specify that the mastermind hires them for a job, while clearly and convincingly stating (while providing convincing evidence) to everyone that J will not be of any help, because she will be asleep (or perhaps go deaf and blind—if that is more convincing). Further, J agrees and they all have a reasonable, strong and justified belief that she will be sleeping (assume this is the case for all involved). Further, J is aware that none of the participants, including the boss, are compelled in any way to go through with their job, or will perform their jobs more effectively, compared to a situation where J was not included (because we want to rule out J’s causal influence on them and on their reasons for proceeding). J has no causal effect on the commencement or fulfillment of the heist whatsoever, or the potential shooting of witnesses. She does not affect the other participants’ motivation for going through with the mission at all. J may have the desire to contribute, but she is entirely and reasonably aware that she will be unable to, as is everyone else. J will be asleep during the heist, and they all know this.

With the above stipulations in mind, and if they now go through with the heist, it is difficult to accept that J has a morally relevant role, or that J is liable to be killed for her “participation”. If we still have some intuition of J’s residual liability after the noted revisions, it is presumably rather that she fails to fulfill an obligation to stop or counteract the heist. E.g. she fails to call the police beforehand while knowing that the heist will commence, and knowing of the potential fate of the witnesses. Here, it is her omission, i.e. unwillingness to fulfill the obligation to help, and potentially save the witnesses which incriminates her. Her omission was the (quasi) cause of the outcome of the witness getting killed, because the outcome is counterfactually dependent on her omission. If she had called the police, the witnesses would have been unharmed (let us assume). Thus it is again causal responsibility which grounds her incrimination.

With the above clarifications, it is no longer clear that it is J’s non-causal participatory intention, or any role established in light of the mere participatory intention, which incriminates her. Indeed, as noted above, it is presumably difficult to even accept that J should be considered to have the role of ‘lookout’ or any role in the heist. The morally incriminatory relevant notion of role is plausibly at the very least in part defined by the effectiveness or expected effectiveness of performing that role to some tangible effect on the project. If we accept that she certainly will be sleeping, and if we ignore her omission of stopping the heist, she for all morally incriminatory intents and purposes could be sleeping in her bed miles away. She is not incriminated in the heist, unless we accept the absurd notion that mere geographical closeness in itself is incriminatory.

But what about the firing squad case? In that case, the individual may have the aim, but she clearly fails to make a causal contribution to the outcome. The actions of the other participants are not clearly reliant on the contribution of the noted ineffective shooter. She is going to misfire, but clearly knowing this, we presumably still find it permissible for the sniper to kill her in order to save the prisoner. Why? Does this suggest that something like a non-causal notion of role can be incriminatory? Not obviously. Our intuitions of incrimination relate to foreseeability and predictability, and how they constitute at least part of the epistemic condition for moral incrimination with regards to actions and “attempts”. The shooter falsely, but reasonably, believes she will contribute to the outcome in some capacity. She has informal background evidence of e.g. the gun’s and cartridge’s ordinary reliability. She does not merely have that aim, indeed she is actually testing her mettle in the clearest way imaginable, i.e. by performing an action (aiming and pulling the trigger) which she believes will contribute to a highly immoral outcome. That is usually sufficient for individualist MR for an attempt. More importantly with regards to group harm, she is causally responsible for posing a threat, i.e. she is causally responsible for bringing about the circumstances in which it is reasonable to believe she is likely to harm someone. Therefore killing her is permissible, when the stakes are sufficiently high, as in this case.

If she on the other hand had pulled her trigger knowing that her rifle was surely going to misfire, i.e. knowing that she would fail to contribute with near absolute certainty, I imagine our intuitions regarding her incrimination would change radically. In that case, it would not be an attempt at all, and she would not have been causally responsible for posing the threat, since that assumes at least predictability of harm. She would have been a bystander who just happened to have the impotent aim of contributing. Though she was holding an ineffective rifle, she could for all morally relevant intents and purposes have been holding a wooden stick. Now, the sniper may still be justified in, or excused for, killing her, because it is reasonable for the sniper to believe she will pose a threat. But, if the sniper knew she was not a participant in the project at all, and not at risk of contributing, the sniper’s justification is not clear. Of course, the case has not been cleared from omissions either. E.g. by volunteering, believing she will be likely to contribute to an innocent person’s death, we can imagine she has failed to take all necessary steps to hinder the harm. I.e. her incrimination may be partially grounded in her causally underpinned omission. We might also imagine that her being in the firing squad would causally lend moral support to the other shooters, which also suggests a causal contribution we need to consciously discount in order to gauge our intuitions of incrimination accurately. Clearly there is room for some causal or quasi-causal notion of contribution in this case, which plausibly drives our intuitions of incrimination. Based on her volunteering to kill an innocent person, we may also believe she is a bad person in general and speculate that she is causally responsible for other unrelated bad things while being associated with the particular group which has orchestrated the murder. This may again affect our intuitions of liability in the current case, so conscious effort has to be expended in order to rule out the associative intuitions from such a character judgment. However, if we actively ignore these in order to rule out causal responsibility for posing a threat, it is difficult to see why she should be incriminated in the harm. This again suggests that assumptions of causal responsibility ground incrimination in the group project, not causally inert participatory intention.

6 Complicity and the responsibility dilemma

The complicity account purports to ground severe incrimination non-causally. However, as I have shown, though it seems to jettison the causal responsibility condition, it inadvertently reintroduces it with its reliance on the participatory intention condition. It is not plausible that acting with a participatory intention is severely incriminatory, unless it implies a causal contribution to the harm in question, or implies a causally underpinned omission. Or, alternatively, it implies causal responsibility for posing a threat. Thus, causal responsibility seems necessary for complicity. If this is so, how does this affect the complicity account’s ability to solve the responsibility dilemma? Recall, the problem was that both combatants and civilians to a large extent fulfilled the individually necessary and jointly sufficient conditions for MR. The proposed solution was to adjust the stringency of the causal responsibility condition in particular. However, regardless of such an adjustment, it would lead to one of two equally problematic scenarios: (1) an insufficient number of combatants are incriminated to make just war feasible. (2) too many civilians are incriminated, putting them on equal moral footing with combatants—i.e. implausibly making both liable to attack.

With the above in mind, it is clear that the complicity account is an inadequate solution to the dilemma. Assume both combatants and civilians fulfill the necessary background conditions for MR, i.e. they are minimally aware of what they are doing, have sufficient mental capacity, and act voluntary. If they also meet the participatory intention condition, they then fulfill the individually necessary and jointly sufficient conditions for complicity. However, since the participatory intention condition is met if the individual is causally responsible, we again end up incriminating combatants and civilians to the same extent. Just as before, adjusting the causal responsibility slider is not going to solve the problem, since a large proportion of combatants and civilians fulfill it to the same extent. The complicity account is not an appropriate tie-breaker.

Bazargan in particular seems to be aware of the problem. Thus, he attempts to introduce an additional safeguard against equal incrimination. Namely, he asserts that degree of incrimination depends on the centrality of the role one plays in the collective project, where only those who play a sufficiently central role, are incriminated to a sufficiently high degree to be liable to be killed. He can thus concede that combatants and civilians to a large extent fulfill the conditions for complicity, since they e.g. satisfy the causal responsibility condition for complicity. However, if only combatants fulfill roles which are sufficiently central, only combatants are liable (by and large). Specifically he writes: “The function of civilians, such as tax-payers, is to contribute in ways that have only a marginal impact, at best, on the war. So even if a typical tax-paying civilian, for example, ultimately contributes no more [or no less] to an unjust war than an ineffective combatant does, the ineffective combatant can still bear substantially greater complicitous liability than the civilian. A combatant’s role, in a straightforward sense, is designed to contribute to a degree far greater than the typical civilian’s” (Bazargan 2013: 189, emphasis mine). Bazargan can thus concede that both combatants and civilians are complicit, because they both have a causally underpinned participatory intention to play a role in the collective project. However, even if they are causal equals, the claim is, that they do not by and large fulfill an equally central role in the “war” project. Particularly, his claim is that combatants by and large tend to fulfill much more central roles compared to civilians. Thus, if he can successfully argue that we should set the role-centrality requirement for liability high enough to include most combatants, and exclude most civilians, he has then solved the dilemma. He has done so, regardless of the fact that the complicity account—contrary to Bazargan’s original claims, and what he attempted to show with the heist case—intimately relies on the incriminated participants being causally responsible.

Unfortunately, Bazargan fails to give a substantive account of role centrality. For a project which ultimately hinges on this notion, which potentially has far reaching revisionary consequences regarding whether or not we are allowed to kill civilians, this is a problem. It leaves us with limited means of examining whether role centrality provides a plausible normative distinction. It seems we are forced to rely mostly on our immediate intuitions of role centrality in order to gauge whether his account is plausible or not. So, do we generally consider combatants to play roles in war, which are designed to contribute to a far greater degree than the typical civilian? The common sense answer here is presumably “yes”. However, the answer to this question is in danger of piggybacking on nothing more than our archetypal distinction between innocent civilians, and soldiers as deadly tools of war. But we have reason to be highly skeptical of these intuitions if we, as Bazargan does, willingly reject them in favour of a theoretical approach, as when we are prepared to reject the common sense notions that both just and unjust combatants are liable to be killed. Further, and more importantly, an obvious interpretation of role centrality is, problematically, that it hinges on some notion of causal contribution. The common sense answer to why combatants’ roles are designed to contribute far greater to the war than civilians, may e.g. be that “they generally tend to make a larger contribution to the war”. But the premise of this whole discussion is that Seth Lazar has provided us with arguments to the effect that this common sense answer is actually wrong. Our assumption is that combatants and civilians to a large extent contribute equally. If this is what grounds our intuitions of role centrality, we should be much less inclined to accept the common sense answer above.

If role centrality is supposed to be based on something else than individual causal contribution, it is unclear how role centrality constitutes a clear distinction between those who are liable to be killed, and those who are not. Bazargan wants us to accept that two people of sound mind, can foreseeable and voluntarily make the same causal contribution to a harm or a threat of harm, but where one of them plays a more central role with regards to that harm than the other, because her role is designed to contribute more. Indeed, one of them may be killed in light of this more central role, if it alleviates the threat in question, while the other may not. Though Bazargan does not give a substantive account of role centrality, he does hint at how this centrality may be justified. E.g.: “a combatant occupying a[…] role of this type contributes to a degree far greater than the average tax-paying civilian […] as is evidenced by the fact that the finances required to train and fund a combat engineer, a boom operator, a navy technician, etc. far outstrips the funds garnered by the government from the average tax-paying citizen“(Bazargan 2013: 189–190). The first thing to note is a seeming conflation of significantly different notions of role centrality measurements. In one case, it is the economic contribution of the person which determines the centrality of the role. In the other, it is the cost of training the person—not her economic contribution—which determines role centrality. Regarding economic contribution, it is unclear how that in itself would e.g. have made a difference to J’s incrimination in the heist case. Should it really matter whether the group leader had spent millions of dollars training J as a lookout, or only a few dollars, if all (including J) know J will still be entirely causally inefficacious regardless? Doubtful. On the other hand, if J actually did contribute, e.g. by foreseeably influencing the other participants in the heist, would we then consider it mitigating with regards to J’s liability, if we later came to know that the leader had only spent a surprisingly limited amount of resources training J? Again, doubtful. The reason presumably is, that “training” is merely a proxy for effectiveness at fulfilling the role, but where ‘effectiveness’ is actually determined by expected causal efficacy. So, when that proxy is screened off by information about the actual circumstances of the case, e.g. J’s expected sleepiness, this naturally takes precedent over the proxy, which then for incriminatory purposes can be bypassed. The actual incriminatory aspect is plausibly J’s actual causal contribution, whether it is direct, based on an omission, or based on J’s causally underpinned threat. Further, in the war case, if we accept Lazar’s arguments (as Bazargan does), we lack reason to expect that the soldier’s training allows for predicting greater causal efficacy than the civilian’s in the war scenario. As far as we believe, they are to a large extent causally responsible for posing the same threat. Thus, if we assume role centrality hinges on causal responsibility, then the dilemmatic incriminatory equality is maintained.

Economic contribution is also a mere proxy. The more economic resources one pours into a project, the more central one’s role plausibly is in that project, but presumably only if these resources plausibly increase the facilitation of the project. If J the lookout had spent millions on a pair of ancient binoculars with broken lenses, J now, in a sense, “contributes” economically to the project. But it does not make his role more central. It fails to increase her moral significance, because this spending fails to increase J’s ability to be causally effective. Economic contribution is a proxy for causal efficacy, a proxy which can be screened off by the actual causal facts if we know them. When we know the causal facts, these facts instead drive our judgments of role centrality and incrimination. Most importantly here, in the actual just war case, we arguably know what the actual or expected causal contribution is. As Lazar argues, that contribution is to a large extent equal, when everything else is screened off. Thus, there is nothing which clearly distinguishes combatants and civilians with regards to role centrality. The responsibility dilemma thus remains unsolved.

7 Conclusion

As shown, there is no clear case for the complicity account’s ability to ground severe individual incrimination non-causally. The prime component and replacement-condition for the causally responsibility condition, i.e. the participatory intention condition, is strongly incriminatory only when it implies causal responsibility. Participation implies either causal responsibility for contributing to the harm, or the threat of harm. The account is thus not a viable alternative to traditional accounts of MR with regards to the noted problem cases of collective involvement, precisely because it succumbs to the same problems. Namely, that in certain cases of collective action, a large proportion of the harm is not sufficiently attributable to seemingly accountable individuals, because the harm was outside their clear sphere of causal influence.

Regarding just war and the responsibility dilemma, the complicity account is an inadequate solution for the reasons outlined above. Firstly, incrimination in a group project based on a causally inert participatory intention is insufficient to allow for attacking and killing people. Alternatively, if we interpret the complicity account as implying causal responsibility, we end up reintroducing the responsibility dilemma, since we have no obvious method for distinguishing between combatants and civilians who are equally causally responsible. There is no clear sense in which a large proportion of civilians are more or less causally responsible for harm or for posing a threat of harm than a large proportion of combatants. Further, the suggestion that role centrality could be a tie-breaker, which would allow for incrimination of combatants only, fails. It fails because role centrality plausibly reintroduces causal responsibility as its criterion, which again, as problematized by Lazar, is approximately equal among a large proportion of members of both groups.