Parfit's Triple Theory
Parfit in seven parts: Part IV
Parfit’s central project in On What Matters is to argue that the best forms of Kantianism, Contractualism, and Rule Consequentialism converge, forming a unified view that he calls the “Triple Theory”. These three theories have traditionally been seen as rivals. Rule Consequentialism directs us to consistently follow whatever rules would be impartially best on the whole (in contrast to the Act Consequentialist view that one should always perform whatever act would be best, even if that involves violating some generally-beneficial rules in exceptional circumstances). Contractualists instead ask what principles everyone could reasonably agree upon, while typically eschewing justifications based on aggregated interests as morally illegitimate, instead allowing only “one-to-one” comparisons in judging who has the strongest moral claim. And Kantian ethics revolves around Kant’s exceptionless Categorical Imperative (in its various formulations).
The strain of Kantianism that Parfit develops is centrally concerned with universalizability. Parfit begins with a version of Kant’s Formula of Universal Law, which he calls the Impossibility Formula: “It is wrong to act on any maxim that could not be a universal law.” One important challenge for this sort of principle is that many acts that can only be exceptional (non-universal) are nonetheless morally innocuous: examples include giving more to charity than the average person, or buying only second-hand books. Other acts, such as refraining from having children, or working in a non-agricultural sector, could (at least briefly) be universalized, but we might not much like the results. Even so, it would be a mistake to object to them on this basis, so long as enough others are happy to do the necessary tasks: just because we need some farmers doesn’t mean that everyone has to do it. These cases are importantly different from the disreputable sort of exception-making (such as free-riding) that we want our moral principles to rule out. The challenge is to identify a principle that excludes just the latter.
Parfit eventually proposes a Moral Belief Formula (MB2) that holds acts to be wrong “unless we could rationally will it to be true that everyone believes such acts to be permitted.” We can happily permit people to give exceptional amounts to charity, or to seek non-agricultural employment. By contrast, we have good reasons not to want everyone to feel at liberty to cheat on their taxes or pollute the environment. So this principle seems helpful for isolating the morally problematic ways of making an exception for oneself.
MB2 may be a helpful principle, but it also raises further questions. For one thing, it presupposes an independent account of the “morally relevant” features of actions. We may often have an intuitive sense of these, but it is less clear whether this sense is sufficiently definite for Parfit’s purposes. Consider: will there always be exactly one “correct” description of an act, which precisely specifies what other acts count as being of the same (moral) type? If it is sometimes indeterminate what falls under the description of “such acts” in MB2, then the principle may sometimes fail to yield a determinate verdict as to whether or not the original act was wrong.
Parfit’s next major revision of Kantian ethics stems from what he calls the High Stakes Objection. Suppose that murdering someone is the only way to save your life. You might then rationally will that everyone believe such egoistic acts permissible, since however bad that result would be for you, it’s probably still better than dying immediately. Parfit’s solution is to switch from asking what the individual could rationally will, to instead ask (in “Contractualist” fashion) what everyone could rationally will, or agree to. This then yields Parfit’s Kantian Contractualist Formula: “Everyone ought to follow the principles whose universal acceptance everyone could rationally will.”
But what if acceptance of a principle had consequences other than the downstream actions it produced? For example, suppose that if everyone believed that reading children’s stories was immoral, then—magically—climate change would be averted. (Crucially, we are to imagine here that actually reading children’s stories remains harmless. We merely have to believe that it’s wrong, we don’t have to act accordingly, in order to secure the benefits.) Why should the usefulness of those beliefs be at all relevant to whether their contents are true? Given the essential gap between general rules and specific circumstances, such a rule-based approach to ethics strikes me as fundamentally misguided.
According to Parfit’s Kantian Rule Consequentialism, “Everyone ought to follow the optimific principles, because these are the only principles that everyone could rationally will to be universal laws.” Let’s put aside our concerns from the previous section, and assume, for sake of argument, that Kantian Contractualism is correct: everyone ought to follow the universally willable principles. Is Parfit right to think that the optimific principles are uniquely universally willable in this way? We can break this down into two further questions: (i) Is it true that everyone could rationally will the optimific principles? (ii) Are there any other principles that everyone could rationally will?
We can probe the first question by considering someone who would be disadvantaged by the (overall) optimific principles: perhaps their life, and the lives of their loved ones, would have to be sacrificed for the greater good. Could someone rationally will such tragedy upon themselves?
Parfit affirms a wide value-based objective view of reasons, according to which “we often have sufficient reasons” to act in either the way that is personally best, or the way that is impartially best. While Parfit lacks decisive arguments for this view, many may nonetheless agree that it is always at least rationally permissible to prefer the impartially best outcome. The more permissive our conception of rationality, the easier it will be to accept this claim, but the more challenging it will prove to exclude competitor principles. After all, why couldn’t some (non-optimific) deontological principle, such as one proscribing killing people as a means, also be universally rationally willable?
When Parfit calls a principle ‘optimific’ or ‘best’, he means these terms in the impartial-reason-implying sense, meaning whatever “from an impartial point of view, everyone would have most reason to want.” Importantly, this could diverge from typical consequentialist evaluations (of what maximizes impartial welfare): some deontologists claim, for example, that we all have impartial reasons to prefer that innocent people not be killed as a means, no matter what else is at stake. This threatens to rob Parfit’s “Kantian Rule Consequentialism” of much of its apparent interest and significance. If it might be deemed “best” to abide by deontic constraints no matter the cost, the resulting view would seem “consequentialist” in name only.
Parfit might respond to such deontologists by invoking his Deontic Beliefs Restriction: when applying a Contractualist formula to determine what is wrong, it would be viciously circular to appeal to deontic beliefs about what is antecedently wrong as one’s basis for rejecting a principle. One must instead appeal only to non-deontic claims, such as claims about harms suffered (or other features of the acts under consideration). Parfit argues that this restriction helps to push Contractualists in a more (welfarist) consequentialist direction. To see why, compare the following two competing principles regarding whether one may save someone’s life by means of destroying another person’s leg:
[A] The Harmful Means Principle: It is wrong to impose such a serious injury on someone as a means of benefiting other people.
[B] The Greater Burden Principle: We are permitted to impose a burden on someone if that is the only way in which someone else can be saved from some much greater burden.
As Parfit notes, many people find the Harmful Means Principle to be highly intuitive. But if we apply a Contractualist formula to decide between the two principles, the Deontic Beliefs Restriction bars us from appealing to such moral intuitions. We must instead ask what non-deontic reasons agents could invoke for choosing between the principles. And now the defender of the Greater Burden Principle would appear to be on firmer ground, as they can appeal to the fact that loss of life is a much greater harm than loss of a limb (a claim that does not presuppose that any particular act is wrong antecedently to applying the Contractualist formula). This puts significant pressure on Contractualists to instead reject the Harmful Means Principle, and more generally to reject deontic constraints against utilitarian sacrifice—harming some in order to benefit others more.
Evaluating the Triple Theory
When the best forms of Kantianism, Contractualism, and Rule Consequentialism are combined, Parfit claims, the result is his Triple Theory: “An act is wrong just when such acts are disallowed by some principle that is optimific, uniquely universally willable, and not reasonably rejectable.” It is worth quoting at length why Parfit believes this result to be so important:
Of our reasons for doubting that there are moral truths, one of the strongest is provided by some kinds of moral disagreement. […] If we and others hold conflicting views, and we have no reason to believe that we are the people who are more likely to be right, that should at least make us doubt our view. It may also give us reasons to doubt that any of us could be right.
It has been widely believed that there are such deep disagreements between Kantians, Contractualists, and Consequentialists. That, I have argued, is not true. These people are climbing the same mountain on different sides.
We can raise both ‘internal’ and ‘external’ objections to this justification for Parfit’s project. Internally: Parfit’s conclusions leave room for plenty of deep moral disagreement. Many influential accounts of morality, from Act Consequentialism to Virtue Ethics, are seemingly left scaling different mountains. Moreover, even if Parfit is right that the three accounts he focuses on coincide in this way, we might still dispute which is the most normatively significant. Are all three components equally essential? Or does one do the fundamental wrong-making work, while the others are merely incidental?
Externally: there’s plenty of room to dispute the claim that moral disagreement is as threatening as Parfit assumes. In ‘Knowing What Matters’, I argue that actual unanimity would gain us little of metaethical import. The deeper issue remains that there are any number of internally-coherent alternative worldviews against which we can muster no non-question-begging argument. Whether those alternative views have actual defenders or not is irrelevant to how troubled we should be by them in principle.
So what it really all comes down to is whether it can be rationally defensible to maintain a view whose foundations might coherently be questioned. But this is something that anyone who is not a radical skeptic must simply make their peace with. Even the foundations of our most commonsensical empirical beliefs (that the external world exists, has existed for more than five minutes, and will continue to exist tomorrow) might coherently be questioned, after all. If a radical skeptic disputes all our starting points, there will be no way to convince them that we are right. But we needn’t be too troubled by such intransigence. What matters is not whether others might disagree with us (of course they might!), but whether they can offer positive reasons for thinking that some alternative view is more likely to be correct than our own. Skeptical doubts may prompt us to closely examine our beliefs, but they do not settle what we should conclude.
This stance is not (objectionably) dogmatic: we should always be open to the possibility of receiving good reasons to revise our views. Disagreement can be relevant, when it’s evidence that we’ve made a mistake by our own lights—a blunder we would disavow upon closer examination. But a fundamental moral disagreement instead reveals that you’ve met an agent who has different moral starting-points from you. That might create practical difficulties, but it is not, by itself, evidence that your reasoning has in any way gone awry.
So I’m dubious of Parfit’s convergence-seeking project. Methodologically, I’d sooner encourage moral theorists to develop the principles they find most plausible, no matter that others might disagree. Moving on to this question of substantive normative judgment, then, here’s my central reason for rejecting Parfit’s Triple Theory:
I find it implausible that our moral evaluation of an action should indirectly follow from our evaluation of some broader rule or principle. In particular, the mere fact that the best uniform (or universal) rules recommend an act does not mean that this specific act is any good—the rules’ benefits may stem from other cases. This prompts a couple of deep challenges to Parfit’s rule-based approach: (i) When an optimal act is ruled out by optimal principles, why prioritize the principles—why should acting optimally ever be considered “unjustifiable”? (ii) Different people might do better to be guided by different principles—so, even on a rule- or principle-based approach, why require uniformity?
Parfit simply doesn’t have any arguments to win over those of us who are more drawn to Act than Rule Consequentialism.
Five Valuable Upshots
While I’m unsatisfied by Parfit’s positive view, some important progress in ethical theory is made in On What Matters (vols 1 and 3). Here are what I take to be the five most important lessons:
As mentioned above, Parfit argues powerfully that the Deontic Beliefs Restriction should move Contractualists in a more consequentialist direction (rejecting deontic side-constraints against instrumental harm).
Parfit offers a compelling objection to the Kantian assumption that utilitarian sacrifice (absent consent) essentially treats the sacrificed party “merely as a means”. In Parfit’s Third Earthquake scenario, we are invited to imagine that you save your child’s life by using another person as a shield, crushing the other’s toe without her consent. But we are to further suppose that you refrain from saving your own life, because saving yourself would have crushed a second of the other person’s toes. Since you value this other person more highly than your own life, you are clearly not treating her merely as a means, despite using her (without her consent) as a means to save your child’s life. This is an important corrective to this common objection to utilitarian sacrifice. It also threatens to trivialize versions of Kantianism that seek to ground all of ethics upon the prohibition against treating anyone merely as a means.
Parfit argues against “commonsense” moralists (and defenders of the Doctrine of Double Effect) who want to specifically prohibit harming as a means, while permitting comparably beneficial harms that are mere side-effects (or “collateral damage”). He shows that the very objections standardly offered against harming as a means apply just as powerfully against harming as a side-effect. Either option equally harms the subject without their consent, for example, or in the case of killing, robs them of the only life they have.
Arguing against Thomson’s principle that it’s permissible to beneficially redirect existing threats (whether bombs or trolleys) but not to introduce new ones, Parfit offers a Fire and Flood scenario in which it would be clearly morally better to save more lives by flooding a burning building, killing one person in the basement, than to merely save a few lives by redirecting the fire into another room where five people would still be killed.
Parfit concludes that we should all accept the following Principle of Unintended Threats: “When there is some unintended threat to people’s lives, such as some fire, flood, approaching asteroid, or runaway train, we could justifiably do whatever would cause fewer people to be killed.” This requires some revision of common verdicts in trolley cases (to permit pushing one in front of the trolley if that would save five). In other controversial cases—such as Transplant: killing one to provide vital organ transplants to five others—Parfit claims that his principle “does not apply, because these cases do not involve unintended threats to people’s lives.” He unfortunately does not explain why organ failure, for example, does not count as an “unintended threat.” Perhaps he means to restrict the principle to external threats? Or perhaps to threats of which a single instance is capable of killing multiple people? More work may be required to pin down the best version of Parfit’s principle.
Non-consequentialists often claim that we have an agent-relative duty to avoid (ourselves) killing, even to prevent more killings by others. But, as Parfit points out, this does not seem to hold with full generality. If we imagine cases involving no wrongdoers, it instead seems that we should all share the same aim of ensuring that as few people as possible are justifiably killed. (For example, in Parfit’s Power Plant case, you and I are the only people who can prevent a nuclear power plant explosion. If you do it, one person would thereby be killed as a side-effect. If I do it, two people will be killed as a side-effect. Clearly, you ought to do it, even though this means that you are responsible for (justifiably) killing one person, which you could have avoided by leaving me to justifiably kill two.)
Parfit thus argues that it’s only in cases involving wrong-doers that our reasons to avoid killing seem agent-relative. And this intuition might be explained away as, e.g., reflecting a useful rule for protecting ourselves from exploitation (by those who might otherwise threaten great harm merely in order to manipulate us).
(For more on the Triple Theory, see sec. 5 of Parfit’s Ethics.)