Intention and the doctrine of double-effect
So far, we have found that the proportionality and victim-centred conditions on so-called necessity killing already drastically limit the scenarios in which abortion is permissible if it amounts to homicide. In fact, as I have said, the relevant conditions are only ever met where killing the fetus can save the life of the pregnant woman but where the fetus itself cannot be saved by any action.
According to some, however, there are yet further constraints on morally justified killing. One of these putative constraints stems from a distinction, often made in moral philosophy, between killings that are directly intended and those which are indirectly intended. To kill a person with direct intention is to kill her either as a means to an end or as an end in itself. To kill with indirect intention is to bring about someone’s death as a known side-effect of some other desired result which could be logically, although maybe not practically, achieved without causing the death. For example, a burglar who shoots and kills an apprehender in order to silence him and eliminate any witnesses kills with direct intention, since the death of the victim is something he is aiming to achieve by shooting. We can contrast this with Glanville Williams’s famous insurance bomber, who places an expensive package on a plane along with a bomb, intending to blow up the plane and collect the insurance on the package. The bomber knows that all of the plane passengers will be killed when the bomb explodes, but he does not aim directly at their deaths, since those deaths are not integral to his fulfilling his purpose. The same distinction is sometimes described as the difference between killing with the intention to kill or with the mere knowledge or foresight that one’s actions will cause death (or what some, differently again, distinguish as ‘direct’ versus ‘oblique’ intent).
Some would argue that there is an absolute prohibition on the directly intentional killing of an innocent person. If this were true, then, on the personhood proviso, most forms of abortion would appear to be prohibited even where the first two conditions on necessity killing—the proportionality constraints and the victim-centred constraints— are met. Moreover, those who read an important moral difference into the difference between direct and indirect intentional killing may attempt to distinguish the violinist case from abortion on this ground, if it were assumed that the other conditions for necessity killing are equally met in both. Is it possible to claim that when unplugging the violinist one only indirectly intends his death, but that all or some abortion directly intends the death of the fetus? If this were true, it may be another route to claiming that abortion is impermissible whereas unplugging the violinist is not.
In the previous chapter, we already encountered some important distinctions between kinds of conduct: that between killing and letting die, and between doing and allowing harm. The distinction between killing directly and indirectly is different from both of these. Indirect intentional killing is not the same as failing to save. As Thomson acknowledges, ‘an indirect killing is perforce a killing’, whereas a man who has never killed may still have failed to save many lives. Thus, the proportionality and victim-centred conditions for the necessity defence to killing apply regardless of whether the killing in question is directly or indirectly intended. One clearly does not have carte blanche to kill whomever so long as one does it with indirect intention.
The difference between directly and indirectly intended killing will only ever be relevant, then, where the other conditions for morally permissible killing are already made out. Where this is so, the key claim is that there are residual side- constraints on the manner in which one or more persons may kill another: whereas killing with indirect intent is sometimes permissible where the gain in terms of life preserved is great enough, directly intended killing is absolutely prohibited and does not ever admit a results-based justification. As we shall see, the implication of this claim for those propounding it is that the causal structure of an abortion might determine whether or not it falls under the absolute prohibition. On John Finnis’s view, an abortion by hysterectomy (the removal of an entire womb and the fetus with it) could conceivably be permissible even if the fetus is a person, but an abortion that crushes the fetus’s skull in order to extract it never could be. Underlying the suggested importance of this distinction is the more general proposition that there are some sorts of acts it is never permissible to perform even despite a great pay-off in terms of life saved, net welfare, or some other calculable good. These are what philosophers call ‘deontological constraints’.
As Finnis points out, Thomson seems to acknowledge in principle the existence of such constraints when she recognizes that there are ‘drastic limits’ to the right to self-defence, such as that one probably may not torture another person even to protect oneself from deadly harm.40 However, she is not convinced that any side-constraints render abortion impermissible merely because fetal death is more directly intended than the death of the violinist in her scenario.
To reiterate, it is not the case that one has permission to kill whomever so long as death is not specifically aimed at but is rather the known side-effect of some ulterior purpose. For the most part, English law does not distinguish the two, and for good reason. The insurance bomber example is a good illustration of why intention to kill in criminal law encompasses both the direct and indirect kinds. Even though the deaths of all of the passengers are, strictly speaking, side-effects without which the bomber’s purpose would still have been accomplished, it would be absurd if this could make a difference to his liability for murder, which requires an intention to kill or cause grievous bodily harm. The general rule in English law, set down by the House of Lords in R v Woollin, is that what is foreseen as a virtually certain consequence of one’s act will almost always be regarded as intended.41
There is an exception to the assimilation of direct and indirect intent in the English law of murder, however. This is the ‘doctrine of double-effect’ (DDE). The typical example of DDE at work in the law concerns the administration of high doses of pain- relieving medication by physicians to patients near the end of life, with the purpose of providing comfort, but in the certain knowledge that they will hasten death. In this circumstance, the law does not impute the intention to kill to the physician. Even 
40 Thomson (n 2) 53.
though death was foreseen as inevitable, harm to the patient is taken to be so at odds with the overarching aim or purpose of the act—bringing the patient relief—that it is altogether struck out as the meaningful intention. Here, at least, the law accepts that the difference between what is foreseen as virtually certain and what is aimed at can matter for the morality of homicide.
With all of this in mind, we can pose two questions about intention-based side-constraints. First, is there an important moral difference between directly and indirectly intended killings, such as would render abortions which aim at the fetus’s death absolutely prohibited? Second, if a moral difference of the kind exists, could the DDE be used to show that most abortion is in fact a species of indirectly rather than directly intended killing, even if it involves a fatal attack on the body of the fetus?