The Problem of the Mistaken Defender
Most accounts of permissible self-defense focus exclusively on articulating conditions under which the need to avert a genuine threat generates a permission to impose harm on another person. As a consequence many of the central cases in the literature of self-defense presuppose that the defender enjoys a sort of localized omniscience; he knows not only that the threat posed is genuine, but also the exact chances of success and attendant side-effects of his various defensive alternatives.
Beyond facilitating hypothetical or ex-post moral evaluations, there are two desiderata for a satisfying theory of the permissibility of imposing defensive harm. First, there are some cases where intuitively an agent is justified in defending herself, despite facing no genuine threat, because she was reasonably mistaken. A satisfying account should explain the intuitive permissibility of such reasonably mistaken self-defense (without over-generating). Second, as Ferzan, Frowe, and Lazar each emphasize, one of the key roles of such a theory is to guide agential deliberation in the moment.[1] This paper argues that contemporary accounts of permissible defense are ill-suited to address these two challenges. Given the importance of these two roles for a full theory of permissible self-defense, I argue that these failings give us good reason to look to supplement our moral theories.
The central cases of reasonably mistaken defense are:
• Incompetent (Culpable Apparent Threat): Defender is being pursued by Incompetent Murderer, who hates Defender and wants him dead. Both Defender and Incompetent Murderer mistakenly believe Incompetent Murderer’s gun to be loaded. [2]
• Joker (Responsible Apparent Threat): Joker decides to pull a practical joke on defender, by pulling a gun Joker knows to be a fake and convincingly pretending to threaten Defender with it.[3]
Neither incompetent murderer nor Joker is intuitively wronged by defender; nor does it seem appropriate to hold that Defender is liable to counter-defensive harm from them. However, merely believing (even on good evidence) that she faces a genuine threat does not ensure that a defender is permitted to impose defensive harm. Just consider
• Merely Reasonable (innocent apparent threat): Defender has been told by someone she trusts that Bystander is an assassin about kill her. When Bystander begins walking toward her, Defender kills him, believing it necessary for self-defense.
Whatever we say about the significance of the defender’s standpoint, we should hold that Defender in this case has wronged Bystander (though perhaps Defender is excused, given her evidence.)
A satisfactory theory must enable epistemically limited agents to anticipate which of their options is morally permissible (be action-guiding), and should permit some reasonable mistakes in a way that explains the significance of the defender’s viewpoint. Contemporary theories fail with respect to one or more of these challenges: either they fail to be action-guiding, or they get the wrong results in incompetent murderer and joker by implying that Defender acts impermissibily or is liable to counter-defensive harm (or both), or they capture the right result, but in an ad-hoc non-explanatory way that is at odds with the most straightforward intuitive gloss of the underlying moral facts. The theories I will focus on can be divided into two camps: pure liability views, and views that treat the defensive permission as a justified infringement of Apparent Aggressor’s rights.
The most popular type of liability account holds that an agent incurs liability in virtue of being responsible in a particular way for an objectively unjust threat. There are two major strains of the objective-threat account. The first (championed by Thomson (1991) and Uniacke (1994)) grounds liability in an agent’s being causally responsible in some way for an unjust threat to another’s rights.[4] The second has come to be known as the ‘Responsibility Account’, and holds that an agent S is liable when he is morally responsible for an unjust threat. This is the view advocated in Otsuka (1994) and McMahan (1994, 2005, 2009).[5] These views either get the wrong result, implying that the defender is liable, or get the right results for the wrong reasons (the defender acts wrongly, but is not liable because she has full epistemic excuse). In either case such views fail to be action-guiding.
Some (including Ferzan (2005) and Frowe (2010)) have aimed to resolve these failures by offering accounts on which an individual can incur an objective liability by being responsible (in the right sort of way) for another’s justified belief that he is objectively liable. Belief-based liability views either build in an external condition (aggressor culpability) and thus lose their claim to action-guidingness because defenders are not in a position to tell whether the condition is met, or ground aggressor liability in the defender’s beliefs, which yields the wrong results in Mere reasonable belief.
Liability accounts share a general flaw: in grounding the permissibility of defensive harm in S’s liability, they either generate permissions that are intuitively too restrictive (failing to permit reasonable mistakes), or else attach apparently disproportionate costs to merely causing someone to falsely believe that they are under threat.
Treating the cases instead as simply justified infringements of Apparent Aggressor’s rights is strained and ad hoc unless supplemented by a more explanatory theory. Unlike genuine defense cases, mistaken defenders are not forced to choose between their own life and the aggressor’s, and it is unlikely that the good of action-guidance is great enough to justify Defender in killling the apparent aggressor. Even if it does deliver this result, there is reason to worry that the account will over-generate, granting defensive permissions in Mere reasonable belief as well. Finally, even if the account can secure the right permission profile, it is non-explanatory: the permission is no longer linked in the right sort of way to anything the apparent aggressor did to become vulnerable to such infringement.
So, if we’re convinced that action-guidingness and the mistaken defender are important things for a theory of defense to accommodate, we need to look beyond these presently available accounts for a theory (or extension) that can play that role.
[1] Ferzan, K. (2005). Justifying self-defense. Law and Philosophy, 24(6):711–749; Frowe, H. (2010). A practical acount of self-defence. Law and Philosophy, 29(3):245–272; Lazar, S. (2012). Necessity in self-defense and war. Philosophy and Public Affairs, 40(1):3–44.
[2]This particular case appears in Frowe 2010; Ferzan 2005, McMahan 2005 and Tadros 2011 discuss very similar cases. (McMahan, J. (2005). The basis of moral liability to defensive killing. Philosophical Issues, 15(1):386–405; Tadros, V. (2011). The Ends of Harm. Oxford University Press.).
[3]Close parallels to this case occur in McMahan ( (2011). Who is liable to be killed in war? Analysis, (71):544–59.) and Ferzan (The bluff. (ms)).
[4] Thomson, J. J. (1991). Self-defense. Philosophy and Public Affairs, 20(4):283–310; Uniacke, S. (1994). Permissible Killing: The Self-defence justification of homicide. Cambridge University Press.
[5] Otsuka, M. (1994). Killing the innocent in self-defense. Philosophy and Public Affairs, 23(1):74–94; McMahan, J. (1994). Self-defense and the problem of the innocent attacker. Ethics, 104:252–290; McMahan, J. (2009). Killing in War. Oxford University Press.