In Defense of Stakes-Sensitivity

By Richard Y Chappell🔸 @ 2025-10-19T21:35 (+16)

This is a linkpost to https://www.goodthoughts.blog/p/in-defense-of-stakes-sensitivity

TL;DR: Stakes-sensitive beneficence is compatible with individually-directed concern (even for future individuals—see fn 2). Also, we can have reasons to optimize that are decisive for good practical reasoning even if we do not call the resulting act a matter of “duty”. Good people are concerned to act well, not just to discharge their duties.

In ‘The Case for Strong Longtermism’, Greaves & MacAskill appeal to what we may call the Stakes-Sensitivity Principle (SSP): “When the axiological stakes are very high, there are no serious side-constraints, and the personal prerogatives are comparatively minor, one ought to choose a near-best option.”

This is an extremely modest principle. Consider that moderate deontology allows that even the most serious side-constraints can be overridden when the stakes are sufficiently high. SSP does not even claim so much. It is compatible with outright absolutism.

Even so, Charlotte Unruh’s ‘Against a Moral Duty to Make the Future Go Best’ disputes this modest principle, which she characterizes as “constrained utilitarianism” (I prefer the term “beneficentrism”, as the idea’s appeal is by no means limited to utilitarians). Unruh writes:

On this view, demands of beneficence always generate moral duties to aid, unless these duties are outweighed by a prerogative.

Deontologists can reject this characterization of beneficence. They can hold, against [SSP], that we have no general duty to make things go best from an impartial perspective. The deontological duty of beneficence is limited in principle, or, as Wiggins puts it, ‘by its true nature’.^[1]

Wiggins, in ‘The Right and the Good and W. D. Ross’s Criticism of Consequentialism’, argues that Ross’s conception of beneficence as a general duty to promote intrinsic value (all else equal) is in tension with Ross’s own insight that “the essential defect of the ‘ideal utilitarian’ theory is that it ignores, or at least does not do full justice to, the highly personal character of duty.” Wiggins suggests, plausibly enough, that it is not just the other Rossian prima facie duties but beneficence too that should be understood as “personal” in character:

The beneficent person is one who helps X, or rescues Y, or promotes this or that cause and does so because each of those thing is in its own way an important and benevolent end. His acts are not directed at simply increasing the net quantity of intrinsic good in the world. (p. 274)

I couldn’t agree more! This is precisely the sort of view I defend in ‘Value Receptacles’ and ‘The Right Wrong-Makers’: the utilitarian’s fundamental reasons (and hence fitting motivations) are not about “promoting value”—that’s just a kind of summary criterion—but rather concern promoting the interests of concrete individuals. Crucially, as I argue in those papers, that’s all perfectly compatible with adjudicating tradeoffs between competing interests by giving each its due weight and optimizing accordingly. Caring equally about each separate individual, and making rational tradeoffs, will naturally result in your preferring and choosing the outcome that can be characterized as “maximizing value”. But that’s quite different (and psychologically distinguishable) from merely caring about “the net quantity of intrinsic good.” To insist that maximizing well-being entails such merely abstract concern rather than caring about concrete individuals is a (now-refuted) ideological prejudice.^[2]

Wiggins continues (p. 275):

We can see it [beneficence] as a schema that generates countless more specific, as Ross would say prima facie, duties most of which arise (often in a supererogatory or non-mandatory manner) from the agent’s historic situation, arise from who he is, arise from who his putative beneficiary is, or arise from what goals he has already committed himself to promoting (say, education or music or whatever).

As indicated above, I’m on board with the “schematic” interpretation of beneficence. But no supporting argument is offered for the limiting “most of which arise…” clause. Should we really think that some “putative beneficiaries” just don’t matter, such that a moral agent has no pro tanto reason whatsoever—even when all else truly is equal—to benefit them or wish them well? Consider what that would mean. Suppose the individual in question were drowning in Singer’s pond. And our agent’s finger were resting on a button that would activate a robot lifeguard to rescue the drowning person. Would it really be decent for our moral agent to lift their finger without pressing the button, and justify themselves by saying, “Well, it’s not as though this would advance the cause of music, which is what I’m personally committed to promoting.”

Surely it is vastly more credible (and decent) to grant that each person’s interests give rise to at least some pro tanto reasons of beneficence—to wish them well, and assist them when doing so is costless (or sufficiently low cost, including consideration of opportunity costs), etc. If you want to add a prerogative to favor causes associated with personal interests to some extent, feel free. But such prerogatives, like moderate constraints, are surely stakes-sensitive. If we modify the previous case to add an alternative button that somehow supports the arts, and stipulate that only the first-pressed button has any effect, some may judge that it is now permissible to let the person drown and support the arts instead. But if we next suppose that the life-saving button does not just save one life, but rather averts an entire genocide (or comparable harms from natural disasters), then it would once again be clearly indecent to refrain from picking the vastly better option.

Let us return to Unruh. She wants to defend an incredibly lax view of beneficence:

As I understand the view that the duty of beneficence is limited, prerogatives do not serve to protect agents from the demands of a general and unlimited duty to do good. Rather, prerogatives allow moral agents to exercise latitude in aspects of their lives that have to do with aiding others.

The difference between these two approaches is clearest when we consider costless opportunities to do (more) good, like my first Button case. I can understand seeing a role for prerogatives to limit our exposure to burdensome demands. I cannot begin to fathom a moral perspective on which we should have “latitude” to gratuitously let others die, or otherwise make the world worse than is effortlessly within our grasp.^[3]

Unruh may^[4] allow that we do have a kind of pro tanto reason to maximize the good (in the schematic sense!), but it is a non-requiring (or merely “justifying”) reason. To understand the significance of the requiring / non-requiring distinction, more needs to be said about what moral ‘requirement’ signifies. Moral philosophers standardly proceed as though it’s clear that everyone in the field uses the term to pick out the same concept, but I think this is far from clear. Further, on what I take to be the best understanding of the concept, it is a kind of normative error to discount the force of a reason just because it is “non-requiring”. That is, whereas it is an implicit presupposition of Unruh’s paper (like many others) that normative inquiry should be focused on questions of “duty”, I think a good person engaging in moral deliberation will set their sights higher than just the absolute minimum that they can get away with. And I take it that SSP is just completely indisputable when concerned with the sense of “ought” that guides this more ambitious form of moral deliberation that good people are inclined to engage in.

That is, it is clearly morally better to do more good rather than less, all else equal, whether or not you want to call this a “duty”. Unruh anticipates such a response:

A final defence of the Stakes Sensitivity Argument is that its conclusion should be understood in a weak sense. Even if we are not required to do what is best, it is still better to do more long-term good than less near-term good. However, if self-sacrifice for the present generation and self-sacrifice for future generations both lie beyond the call of duty, it is unclear exactly what difference future stakes make. Saving resources for the future seems to become one of many ways in which we can discharge our imperfect duty of beneficence.

I find this passage very puzzling. The point of Greaves & MacAskill’s argument is to support deontic strong longtermism, i.e. that “in the most important decision situations facing agents today, one ought to choose an option that is near-best for the far future.” This is, to many, a surprising and highly practically significant result, even if the ‘ought’ in question is the ought of ideal practical reason rather than the ought of duty. Many assume that we have more reason to favor near-term interests, and I take the stakes-sensitivity argument (combined with axiological strong longtermism) to show that this common assumption is mistaken.

Longtermism is not just “one of many ways in which we can discharge our imperfect duty of beneficence,” because discharging our imperfect duty is not all that matters, and not what would guide a virtuous person in their practical deliberations in high-stakes contexts. A virtuous person has greater moral ambition and scrupulosity (as do a great many ordinary people who are, in this respect, at least more virtuous than those who lack this aspiration). They are concerned to make a morally near-best decision in high-stakes contexts, e.g. regarding how to direct their philanthropic resources, and this is a very good and fitting concern to have! Crucially, this is not just the narrow question of what is axiologically best, since that might neglect some important non-consequentialist considerations (if you believe in such things). Rather, it addresses the theory-neutral question of moral ambition: what is truly choice-worthy, all things considered with due weight.

1. ^{^}

   I think Unruh’s quotation of Wiggins here is misleading. Wiggins revises Ross’s conception of beneficence in two ways, which he explicitly separates on p. 276. The first is personalizing it, as I discuss in the main post, to make it “not the same as the requirement to ‘make the world a better place’.” This is the revision that is relevant to Unruh’s purposes. The second revision is to claim that “Beneficence is by its true nature restricted,” by which he means that we do not have outweighed reasons but rather no reasons of beneficence at all to engage in (long-term optimal) atrocities: Wiggins’s reasons of beneficence are, in effect, conditional on not violating prior duties. But since SSP is limited to cases in which there are no serious side-constraints, it is transparently compatible with this “restricted” conception of beneficence.

2. ^{^}

   I think this is true even regarding future persons. Quick proof: there is nothing objectionably “abstract” about our reasons of non-maleficence to avoid bringing miserable lives into existence. When we successfully act on such a reason, we may do so “for the sake of the possible future person in question” (in the relevant, perhaps somewhat loose, sense), even though our success means that the “person in question” never actually exists. So, in exactly the same way, there is no principled barrier to our likewise having reasons of beneficence that speak in favor of bringing happy lives into existence, for the sake (in the relevant, perhaps somewhat loose, sense) of those very individuals.

3. ^{^}

   Unruh expands: “On the view that duties of beneficence are imperfect, agents should aid others often enough. However, there is no general duty to choose an option that is best for others, even when choosing this option is not particularly costly for agents.” Again, if you replace “not particularly costly” with “literally costless in every way”, it better brings out the in-principle irrationality of the view.

4. ^{^}

   Though at one point she asserts that “the strength of justifying reasons to aid plausibly depends on factors such as our commitments, intentions, and relationships to present and future people, in addition to or instead of the number of interests at stake.” If we opt for the more moderate “in addition to” (rather than the implausibly extreme “instead of”) reading, then I think what Unruh is pointing to here are just further reasons—stemming from special relationships and such—which can be understood as distinct from (and in no way undermining of) our stakes-sensitive impartial reasons of beneficence.