Collective Blameworthiness and Shared Intentional Action

So far, we have argued that agents can be collectively responsible for an action or an outcome. We have also shown that they can be collectively responsible for an outcome even if it is not the result of their shared intentional action.

It is sometimes assumed that an outcome or action is associated with a fixed amount of blameworthiness such that it would have to be divided and distributed in cases of collective blameworthiness (see e.g. Cohen 1981: 73). However, there is good reason to reject such a “pie model” of collective blameworthiness. If doing arithmetic with different agents’ blameworthiness makes any sense, then the blameworthiness should be multiplied rather than divided in cases of collective blameworthiness. In other words, the involvement of others’ agency in bringing about a bad outcome does not itself decrease or dilute—nor increase or concentrate—an agent’s blame­worthiness for bringing about that outcome.¹⁰ Consider a case where I intentionally bring about a bad outcome together with other agents. Now, replace those other agents with non-agential contributing causal factors (say, advanced robotic devices), but keep my intention and contribu­tion fixed. There seems to be no reason why I would be any less or more blameworthy in the former case than in the latter just because some of the contributing causal factors happen to be other intentional agents rather than advanced robotic devices (for this argumentative strategy, see Mellema 1985: 182-183; Zimmerman 1985: 116-117; Sverdlik 1987: 72).

So, the fact that other agents are causally involved does not in itself diminish or increase an agent’s blameworthiness. It could still be, however, that the fact that individuals act on a shared intention to do wrong is significant for their degree of blameworthiness. Doing the wrong thing together rather than in parallel somehow seems to make it worse. At the same time, however, this intuition is not very strong. Some people we have presented our contrast cases to share our intuition that you and I are more blameworthy in Shared robbery than in Parallel robberies. Others, however, have had no such intuition. Furthermore, even if there were a clear uncontrover- sial intuitive difference in blameworthiness between these particular contrast cases, theoretical arguments would still be needed for showing that a shared intention contributes to a greater blameworthiness for wrongdoing in other contrast cases.

There are at least two ways in which one might try to support the intuition. The first focuses on a causal difference between shared and parallel action, the second on a volitional diffe­rence. The causal difference is that, due to the fact that individual intentions interlock, shared intentions cause actions in a less sensitive (that is, more robust) manner. We argue, however, that

this in itself is not significant, because moral blameworthiness always implicates the will: the degree to which an agent is blameworthy depends on the quality of will she displays in her actions, and the quality of the agent’s will is partly reflected in the intentions that an agent forms, retains and acts on. We therefore go on to consider the idea that the fact that agents have formed a shared intention to do something wrong normally reveals something distinctive about the quality of their wills. The idea is that there is also a volitional difference between shared and parallel action: in the case of shared actions the agents are mutually implicated in each other’s will in a distinct way.

Before considering the two arguments in the sub-sections that follow, a potential methodo­logical worry should be addressed. One might think that it is possible to eliminate any proposed blameworthiness-relevant difference between cases such as Shared robbery and Parallel robberies by removing ingredients from the former case or by adding ingredients to the latter case. It is not obvious that such manipulation would undermine the contrast between shared inten­tional wrongdoing and bringing about the same result by strategic interaction. For example, our robbery in Shared robbery could arguably be jointly intentional even if we lacked common knowledge that we had the intentions required by Bratman’s Intention condition (see Blomberg 2016a). As we have noted, Bratman’s conditions are presented as jointly sufficient conditions, not also as individually necessary. Or we might add to the Parallel robberies case that each of us has a general disposition to be helpful to other robbers, say. Or add that each intends that the other’s intention be effective, but retain the feature that there is no common end. This will be possible given any reductive account such as Bratman’s, which is constructed out of multiple components (a common end, interlocking intentions, common knowledge that there is the common end and the interlocking intentions, etc.). The worry then is how we in the end will know that the blameworthy-relevant feature that we focus on is one that is always present in cases of shared intentional wrongdoing but never in cases of parallel wrongdoing.

The argument that we put forward in section 10.4.2, however, turns on Bratman’s Intention condition, which, as we have noted, is arguably a necessary condition in Bratman’s set ofjointly sufficient conditions. It may be true though, that acting together is not a unified phenomenon, so perhaps our argument only applies to a certain kind of shared intentional wrongdoing.¹¹ Furthermore, it is possible that shared intentional wrongdoing of this particular kind is multiply realizable (Bratman 2014: 36).

10.4.1 Shared Intention, Insensitive Causation and Predictive Significance

Suppose that, in one case, an innocent person is shot and killed by a skilled sniper, whereas, in another case, the shooter is a clumsy and unskilled marksman who hits and kills the target anyway. We can imagine the marksman appropriately saying “But I didn’t think I’d succeed!” when blamed—an excuse that would be unavailable to the sniper. Now, one difference between the two cases is that the sensitivity of the causal pathway between the marksman’s intention to shoot and kill the victim and the intended outcome is much higher than that of the causal pathway between the sniper’s corresponding intention and the intended outcome. C₁ causes E₁ less sensitively—or more insensitively/robustly—than C₂ causes E₂ if the range of nearby counterfactual circumstances in which C₁ would cause E₁ is wider than that in which C₂ would cause E₂ (Lewis 1986; Woodward 2006). So, one might think that less sensitive causation gener­ally increases blameworthiness. Arguably this idea has some intuitive pull at first glance.

This intuition can be bolstered by considering the fact that, compared to the marksman’s intention, that of the sniper has higher “predictive significance” with respect to the intended outcome (Scanlon 2008: 13; see also Foot [1967] 1978: 23—24). One might think that this makes the sniper’s action more wrongful than the marksman’s, thus indirectly increasing the former’s blameworthiness (Scanlon 2008: 31—32, 41—43). Suppose you are in a position to hire either the sniper or the marksman to indirectly kill the target.

Chiara Lepora and Robert Goodin embrace a similar idea when they argue that a higher degree of prospectively known causal insensitivity between a secondary agent’s act of com­plicity and the wrongdoing coming about increases his blameworthiness for the complicit act (2013: 66—68, 102—112).¹² They take such causal insensitivity to be an independent factor that increases blameworthiness directly, not a factor that indirectly increases blameworthiness by making the act of complicity more wrongful. Other things being equal, the wider the range of counterfactual circumstances in which the secondary agent knows that his contribution would make a difference to whether the wrongdoing comes about or not, the more blameworthy the secondary agent is.¹³ Thus, it is the insensitivity that the secondary agent was prospectively aware of that matters: “[Morality] evaluates our action retrospectively only in light of what could and should have been expected to occur, at the time we had to act” (Lepora & Goodin 2013: 68). Hence, according to Lepora and Goodin, the predictive significance of an agent’s intentions or actions is directly relevant to blameworthiness. While they explicitly restrict this proposal to the blameworthiness of secondary agents in cases of complicity, there is no reason why it wouldn’t also apply to the blameworthiness of “co-principals” in shared intentional wrongdoing.

Other things being equal, a shared intentional action will cause the (jointly) intended result in a more insensitive manner than a parallel activity will cause the (severally) intended result.

On Bratman’s account, we have common knowledge of our intentions when we have a shared intention. Hence, we will prospectively be aware in Shared robbery that the probability that we succeed is relatively high. We are aware that, if we succeed, then the robbery will have been brought about with a relatively high degree of insensitivity. Our shared intention in Shared robbery will have higher predictive significance than our strategic intentions in Parallel robberies. Hence, if the bank tellers and the customers were to discover in advance what our intentions were, then they would think their situation worse in the shared case than in the parallel case.¹⁴

On the proposal under consideration, our shared intention not only makes the jointly produced outcome worse from the victims’ point of view, each of us is also more blameworthy in Shared robbery than in Parallel robberies, both for the robbery of the bank and for the robbery of the customers. In particular, each is more blameworthy even for the outcome that we indi­vidually intended to bring about in Parallel robberies. This is because we would know that our shared intention in Shared robbery would cause that outcome under a range of nearby counter- factual circumstances that is wider than that under which our strategic intentions would cause that outcome in Parallel robberies.

Thus far, the claim has been that what is doing the work is the known insensitivity of the causal relation. It is far from obvious, however, that this is the case. Moral responsibility and blameworthiness are distinct from causal responsibility in that they implicate the will. As it turns out, the intuitions at issue can be better explained in terms of motives rather than causes. Consider first the case where you hire either a skilled sniper or a clumsy marksman to assassinate an innocent person. Arguably, it seems morally worse to hire the sniper because this would nor­mally reflect a worse quality of will than in hiring the marksman. In other words, an insensitive causal pathway between intention and outcome will typically be (defeasible) evidence for a worse quality of will. To illustrate, you would normally have to pay a higher cost for the services of the sniper, which would suggest stronger determination on your part. If his services were offered at the same price, then hiring the marksman would suggest some ambivalence on your part, as if part of you wanted the shot to miss. Such a quality of will explanation undermines the motivation for thinking that the known insensitivity of causation is itself relevant to wrongful­ness or blameworthiness.

Now, focus only on the role of the sniper/marksman and the victim. If the sniper’s and the marksman’s desire and intent to hit and kill the target are the same, then the insensitivity of the causal pathway to the victim’s being killed arguably does not itself make a difference to blame­worthiness.¹⁵ The clumsy marksman’s “But I didn’t think I’d succeed!” would not function as an excuse if it were merely a comment about his own view of his probability of success. To be taken as an excuse, it would have to be pragmatically interpreted as a claim that he didn’t actually want or intend to hit and kill (that is, a claim directly relevant to what the quality of his will was). Indeed, this seems to be the commonsense view. Results from experimental phil­osophy suggest that ordinary people’s judgments about blameworthiness are not sensitive to the skill level of a shooter who intentionally kills an innocent victim (see Sousa & Holbrook 2010; Sousa et al. 2015).

All this also applies in cases of shared intentional wrongdoing. We are not more blameworthy for the bad outcome in Shared robbery than in Parallel robberies in virtue of the fact that the causal pathway from our intentions to the outcome is more insensitive in the former case than in the latter. Nor are we more blameworthy in virtue of our prospective awareness of this causal insensitivity. Upon being blamed by a third person for robbing the bank in Parallel robberies, I wouldn’t be giving a valid (partial) excuse if I pointed out that I might not have succeeded if you had collected the jewelry and smartphones quickly and left early.

One might argue that it is different with respect to the foreseen but unintended side-effect that I bring about in Parallel robberies. If I am blamed for bringing it about that you robbed the customers, then I might be giving a valid partial excuse if I pointed out that you might not have succeeded in robbing them if I had been really quick in robbing the bank. But again, this is because (and only insofar as) it is evidence of a quality of will that manifested itself in my behavior. There is such a difference: Other things being equal, one manifests a worse quality of will if one acts on an intention to bring about an outcome that one knows to be bad than if one merely acts with the awareness that the bad outcome is likely to be brought about as an unintended side-effect of what one intends to do. Of course, if other aspects of my will varied between the cases, then this might not be true (for example, suppose that I strongly desire that you successfully rob the customers in Parallel robberies but not in Shared robbery).

This does not mean that insensitivity of causation is completely irrelevant to whether agents are collectively blameworthy for a bad outcome. Plausibly, for the bad quality of the agents’ wills to be relevant to their collective blameworthiness for the outcome, the quality of these wills has to be such that it plays a role in normally explaining why such bad outcomes come about (see e.g. Bjornsson & Persson 2012). It is plausible that this requires that some threshold of insensi­tivity of causation is passed. However, this does not mean that the difference in the sensitivity of the causal process in contrast cases such as Shared robbery and Parallel robberies explains a diffe­rence in degree of collective blameworthiness.

10.4.2 Shared Intention and Quality of Will

As we have noted, the intentions that an agent is acting on at least reflects the quality of her will. As Bratman points out, the fact that action is carried out as part of some larger plan “might constitute or indicate a deeper level of commitment to the action,” and this at least partly explains the concern with premeditation in criminal law (1997: 31—32). The fact that a person commits to and plans to bring about an outcome is in most circumstances at least strong evidence that she cares about and desires that this outcome be brought about. Thus, when a morally bad outcome is intentionally brought about, then the fact that the agent had a prior intention to bring it about normally indicates something about her quality of will which makes her more blameworthy than if she did not have the intention. This explains why, in Parallel robberies, I would normally be somewhat more blameworthy than you for the robbery of the bank while you would be somewhat more blameworthy than I for the robbery of the customers.

However, forming an intention to do something wrong may not merely reflect a deeper underlying commitment to wrongdoing, the forming of the intention can also constitute an additional wrongdoing. Consider the case of a single agent who performs a morally bad action. She can perform it spontaneously in a way that is isolated from any prior activity or plan. She might immediately and spontaneously reveal an intimate secret about an acquaintance to others out of spite, for example. Alternatively, the telling of the secret might be the result of prior commitment and planning. Suppose the spiteful person committed and planned ahead of time exactly when, to whom and how to reveal the secret. Here, the spiteful pre-committed planner has arguably made herself complicit in her own wrongdoing. She not only intends to perform an action that is morally bad, but she also formed and retained an intention that her future self’s intention appropriately brings about and coordinates this morally bad action. Normally, this would make her more blameworthy for revealing the secret. She will be blameworthy for it both in virtue of intentionally performing the action and more blameworthy for it in virtue of earlier forming the intention to perform it and retaining that intention. Now, this is not invariably the case, since forming an intention to do something bad might involve ruling out an alternative course of action that is even worse. Furthermore, sometimes intentions are passively acquired and immediately executed rather than formed prior to the time of action (Mele & Moser 1994: 45). But in most contexts where the agent formed an intention to do something bad, this makes her more blameworthy than she would be if she did the bad thing without having formed the prior intention.

A natural thought is that what is special about shared intentional wrongdoing is that, by ana­logy, there is a shared will in which all the participants are mutually implicated, normally leading to a greater degree of blameworthiness than in otherwise similar cases of parallel wrongdoing.¹⁶ First, that I go out of my way to form a shared intention with you in a case such as Shared robbery would normally reflect a deeper concern for the banks and customers being robbed or a stronger desire for us to succeed than I would have in a case such as Parallel robberies. Hence, the presence of a shared intention normally reflects underlying desires, concerns or value judgments that are relevant for blameworthiness in cases of wrongdoing. Each of us would thus normally be more deeply committed to the intentional achievement of the morally bad outcome in Shared robbery than in Parallel robberies, where this includes both the outcome that the bank is robbed and the outcome that the customers are robbed. Second, recall that the intentions of participants are bound together in shared intentional wrongdoing in a special way. It is not only that their intentions converge on a morally bad common end. On Bratman’s account, each also intends that both their own and the others’ intentions to bring about that common end are effective.

Due to the way the participants’ intentions are interlocked, each is implicated in the wrong­doing not only due to the fact that their own intention is aimed at the bad outcome, but also due to the fact that they formed intentions aimed at the satisfaction of the other participants’ intentions to bring about the bad outcome. When these interlocking intentions are formed, retained and successfully executed, each will be complicit in the other’s wrongdoing as well as responsible for their own contribution to their own intended wrongdoing. Each of the participants in shared intentional wrongdoing not only brings about the intended bad out­come, each also brings it about that their own and the others’ intentions to bring about this bad outcome are formed, retained and successfully executed. When we each do our part of jointly intentionally robbing the bank and the customers in Shared robbery, each does this in the awareness that we are supporting and satisfying the other’s intention that our own inten­tion effectively brings about the morally bad common end. In addition, we are each indirectly supporting our own intention that the other’s intention effectively brings about that same end. Crucially, each of us formed intentions that two intentions rather than merely one intention be directed toward the morally bad outcome and successfully executed. Each is furthermore argu­ably aware that the morally bad outcome is such that the intention of each of us is directed to it (Blomberg 2016b). In Parallel robberies, each merely intends that one intention, our own, suc­cessfully brings about our own intended bad outcome. As in the individual case, this normally (but not invariably) reflects a worse quality of will, which in most circumstances will make participants more blameworthy than they would be in a contrast case where they bring about the same bad outcome as a result of strategic interaction.

There is one final question we need to address. Thus far, we have focused on Bratman’s reduc­tive account of acting together. But does our argument generalize to non-reductive accounts? While we cannot discuss this in depth here, it seems plausible that it is compatible with accounts of shared intention such as Margaret Gilbert’s (1990, 2008) and Raimo Tuomela’s (2007). On their accounts, shared intention involves an irreducibly joint or collective commitment that glues group members together with respect to some joint action. As Gilbert puts it, when sev­eral agents have a shared intention to bring about some goal, then there is “a pool of wills which is dedicated, as one, to that goal” (1990: 7).

If there is a joint or collective commitment to do something morally bad then this is nor­mally an important fact that reveals something about the quality of the wills of the parties to this joint or collective commitment. Furthermore, insofar as the joint or collective commitment is formed by parties rather than passively acquired, forming it constitutes a further wrongdoing in virtue of which the parties become complicit in the morally bad action that ensues. Arguably, the parties will then normally be blameworthy not only due to their individual commitments to do their parts, but also to the way they pooled or united their individual commitments in a joint or collective commitment. Thus, our argument generalizes from Bratman’s reductive account to non-reductive accounts of shared intentions.

10.5