Understanding PDE

Riisfeldt formulates PDE as follows: ‘an action with two effects—one good and one bad—is permissible if and only if:

1. The action is not bad in itself.

2. Only the good effect is intended, that is, the bad effect is merely foreseen.

3. The bad effect is not the means to the good effect, that is, the bad effect is coincidental.

4. The good effect outweighs the bad effect’. (127)

Riisfeldt first attributes PDE to medieval philosopher Thomas Aquinas, to whom it is often credited in virtue of his ethical analysis of actions, particularly killing in self-defence. Aquinas argues that acts are ethically acceptable only if they are acceptable in object (kind of act), intention and circumstances. Applying this general analysis, Aquinas contends that it is permissible to defend oneself from attack (object), even anticipating the aggressor’s death (circumstance: consequence), but only if a moderate defence (circumstance: manner) is used and the aggressor’s death is outside the defender’s intention.2

Later, Riisfeldt anachronistically describes PDE as an ‘attempt to combine’ the ‘fundamentally incompatible’ ethical theories of ‘Kantian deontology and Millian consequentialism’ (128), both originating centuries after Aquinas. But the incompatibility of Kant’s and Mill’s theories is no reflection on Aquinas’s—or on PDE, formulated more coherently (and less anachronistically) below:

1. The act must not be bad in itself (ie, it must be good or neutral, considered independently of its causing the bad effect).

2. The bad effect must not be intended as the end or goal of the act.

3. The bad effect must not be intended as a means to the good effect.

4. The agent must have a proportionately serious moral reason for performing the act.3

The differences between this formulation and Riisfeldt’s are subtle but important. First, note that there is nothing especially Kantian about an act’s being ‘bad in itself’: the first condition simply acknowledges that an act could be ethically problematic for reasons other than its causing the bad effect at issue, the range of such reasons to be determined in the context of the broader ethical theory in which PDE is applied. (So although the condition is non-consequentialist, clearly there are non-consequentialist theories other than Kant’s. In a classical modern formulation of PDE, for example, Gury understood the condition to mean that ‘the author of the action would not be obliged by any virtue to omit the action’4—thus, on this interpretation, PDE would forbid acts that were unjust, cowardly and so on, regardless of whether they also violated Kant’s categorical imperative.) The third condition does not insist that the bad effect be ‘coincidental’—it can be a normal and entirely expected outcome of one’s action without being a means to the good effect. (For example, when I wash dishes I foresee that my fingers will become pruny from long submersion in water, but I do not intend their pruniness as a means to clean dishes.) Finally, the fourth condition is not at all Millian. Rather than focusing exclusively on the pleasure and pain produced by the proposed act, it addresses the proportionate seriousness of the agent’s reason for acting: the agent must have at least as serious a reason—again, as determined in the context of a fuller ethical theory—to pursue the good act and/or effect as to avoid the bad effect. (For example, suppose I can use my last US$20 to feed my children or to feed twice as many children in developing countries via Oxfam. Even assuming my feeding my own children fails to maximise utility, one could easily argue that I have a more serious reason to feed them than to feed other people’s children.) With this corrected understanding of PDE, we can address Riisfeldt’s concerns about its application to cases of palliative opioid and sedative use.

PDE and end-of-life care

First, regarding the requirement that an act not be bad in itself, Riisfeldt questions whether this criterion applies only to tout court (without qualification) bad actions or also to pro tanto (to an extent) ones; he suggests that opioid and sedative use may, by their intrinsic natures, fall into the latter category—and thus be prohibited under some interpretations of PDE regardless of whether they hasten death. A corrected understanding of the principle’s first condition as forbidding acts that are ethically unacceptable for reasons other than their effects readily addresses this concern. Routine and ethically unproblematic use of opioids and sedatives in medical care suffices to show that administering them is not bad in itself—which suggests that pro tanto bad actions are not necessarily forbidden by PDE.

Second, Riisfeldt suggests that continuous deep palliative sedation (CDPS) is tout court intrinsically bad because, from the patient’s perspective, ‘being permanently sedated to unconsciousness is indistinguishable from being killed’ (129). So, he reasons, ‘if killing is bad without qualification then it is reasonable to say that CDPS is as well’ (129). To address this concern, we must distinguish between acts and their effects. Although PDE’s proponents tend to regard death as a bad effect, they do not (as Aquinas’s analysis of self-defence indicates) consider every act that causes it to be intrinsically bad. So even if ‘permanent unconsciousness’ accurately describes an effect of sedative administration (rather than, eg, each dose’s effects being separate and temporary, or ‘unconsciousness’ being the effect and ‘permanent’ a description of anticipated circumstances), and even if we consider this effect as bad as death (which, eg, Boyle disputes5), we must ask whether the decision-makers in cases of CDPS must intend permanent unconsciousness as means to symptom relief or may simply tolerate it as a side effect of appropriate treatment (ie, administering sedatives) for otherwise-refractory symptoms.6 Although the physician pursuing CDPS clearly intends unconsciousness, s/he need not intend permanent unconsciousness.7

Third, Riisfeldt contends that assuming opioid and sedative administration at the end of life can hasten death, ‘death itself can be considered a means of relieving physical suffering’—he terms it an ‘indirect means’ (129–130). This result, he argues, is problematic for application of PDE, whose ‘means criterion has not been crafted to accommodate situations of overdetermination’ (130). However, understanding this criterion as requiring that the agent not intend the bad effect as a means easily resolves this worry: an agent intending only the direct pain-relieving effects of appropriate medications need not intend death or its ‘indirect’ contribution to the same end. Bennett gives a helpful example of this sort: aiming to keep X (who is brilliant) happy in his job, A promotes him instead of Y (who is competent)—which, as A foresaw but did not intend, so infuriates Y that he quits, further contributing to X’s job satisfaction.8 Similarly, in palliative care, not every causal contributor need be intended as an agent’s means.

Finally, Riisfeldt claims that fulfilling the proportionality criterion of PDE requires sacrificing common beliefs regarding the sanctity of human life. This criticism also rests on a misunderstanding—of both PDE’s fourth condition and beliefs about the sanctity of human life. First, PDE requires a proportionately serious moral reason for acting; it does not simply require that one effect ‘trumps’ another. Second, defenders of the sanctity of human life tend to argue that human life may never be intentionally violated—not that no action hastening death, even as a side-effect, may ever be performed. It is this very understanding of the sanctity of human life that often motivates appeals to PDE in difficult circumstances—those in which death (or another significant harm) is inevitable.

Some remaining issues

Although most of Riisfeldt’s concerns can be addressed with a corrected understanding of PDE, some controversies and ambiguities remain. Ambiguities surrounding application of PDE, such as whether or not ‘permanent unconsciousness’ counts as an ethically significant bad effect, tend to be unavoidable flip-sides of its virtues: as an ethical principle, PDE must be abstract enough to apply to a variety of situations—thus necessitating further analysis for application to concrete cases; and as a principle rather than a complete theory, it relies on the rest of its user’s (defensible and compatible) theory for interpretation. Although Kant’s and Mill’s theories are not good conceptual or historical candidates for theoretical contexts of PDE, it would take a much longer essay to survey and assess more promising ones. (However, I agree with Riisfeldt that both intrinsic features of actions and their outcomes are essential to ethical evaluation and ‘somehow need to be reconciled’ (128).)

Finally, even on a corrected understanding of PDE, it undoubtedly will remain controversial in cases such as ‘The Loop’.9 Unless intending ‘causing someone’s death’ is separable from intending ‘stopping a moving trolley with someone’s body’ (perhaps a reasonable distinction if, eg, a soldier can throw himself on a live grenade to protect comrades while not intending his own death), it appears that PDE would forbid redirecting the trolley towards one trapped person to stop it from hitting five others on the same loop of track—which seems counterintuitive to some. Thus, a PDE-based analysis requires a strong commitment to not intending an ethically significant bad effect as a means to a benefit—perhaps, in some cases, stronger than one’s commitment to upholding initial intuitions.