Sacrifice dilemmas are philosophical thought experiments that involve a choice between causing a death and upholding some moral good. The most common sacrifice dilemma is the trolley problem, which in its basic formulation asks whether one should sacrifice one human life to save five. The trolley problem has been used widely in analytic philosophy for half a century, and more recently has found a place as a tool in field of moral psychology. The trolley problem plays a major role in the research of Joshua Greene (e.g. his seminal paper An fMRI Investigation of Emotional Engagement in Moral Judgment). It has become common in the psychological literature to view the propensity to sacrifice as “utilitarian” and the contrasting propensity to refrain as “deontological” (see for example this paper about the trolley problem and autonomous driving).
Philosopher Guy Kahane wrote a series of nice papers on the relationship between sacrifice dilemmas and utilitarianism. He suggests that while trolleyology can tell us some interesting things about “common-sense” moral decision-making, it does not tell us much of anything about how utilitarian or deontological people are.
Trolley Trouble
In Sidetracked by trolleys: Why sacrificial moral dilemmas tell us little (or nothing) about utilitarian judgment, Kahane surveys the psychological literature to build a case for skepticism of sacrifice dilemmas. The first issue is that sacrifice dilemmas are quite odd:
This research focus is rather puzzling. These hypothetical dilemmas are complex, far-fetched, and often convoluted. It would be strange to think that they offer the key to understanding moral judgment in general. If we wanted to identify the building blocks of moral judgment, it would presumably be more sensible to start by investigating simple instances of moral judgment such as the judgment that a malicious lie or bullying violence are wrong, and—giving special focus to developmental questions—work our way up from there. Eventually, we are likely to arrive at special cases where it can seem that lying and violence could nevertheless be permitted (whether because needed to prevent an even greater harm or for some other reason). Sacrificial dilemmas would thus be just a minor (if interesting) branch within a much broader inquiry, and their interpretation would depend on prior groundwork done on much simpler, more basic cases.
Sacrifice dilemmas do have a certain clarity about them, as they are typically designed to pose specific philosophical conundrums as neatly as possible. They also seem to be pedagogically useful as this Socratic dialogue in Michael Sandel’s freshman course on Justice shows. However, it is unclear whether this clarity survives outside of their original context.
The debate between utilitarians and their opponents has indeed often appealed to elaborate thought experiments and fanciful examples, both to criticize utilitarianism and to support it—thought experiments involving, for instance, archbishops and chambermaids in a burning building (Godwin, 1793/1926), the moral integrity of a chemist (Williams, 1973), a child drowning in a pond (Singer, 1972), or a rich uncle drowning in a bathtub (Rachels, 1975). But dilemmas involving runaway trolleys do not figure very prominently in this debate. They were first introduced, and most heavily discussed, as a problem within a strand of deontological ethics (Foot, 1967; Kamm, 2007; Thomson, 1985). To the extent that the aim of this recent empirical research on moral dilemmas is to use the hypothetical cases that most sharply divide utilitarians and their opponents, then this research may be focusing on the wrong examples.
This latter point is worth unpacking more. Philippa Foot introduced the trolley problem in her paper The Problem of Abortion and the Doctrine of the Double Effect, which gives a limited defense of abortion that is grounded in the Doctrine of Double Effect (viz. that there is an ethically relevant distinction between the effects of an action that follow from your direct intention, and the effects of an action that you merely foresee; and that moreover when this distinction is present in a dilemma, it is permissible to resort to a consequentialist calculus). DDE originates in Catholic natural theology (here is Thomas Aquinas using the principle to justify homicidal self-defense), and Foot redeploys this idea to argue that in the cases where a pregnant mother’s life is in danger, abortion involves foreseeing but not directly intending the death of the fetus. The trolley dilemma in that paper involves a bunch of underwater cavers being blocked from safety by a fat man stuck in the mouth of a cave, and we have a stick of dynamite that we can choose to toss at him or not as it were, depending on how our ensuing dialogue on the nature of DDE goes. Similarly, Judith Jarvis Thomson’s A Defense of Abortion introduced the famous violinist sacrifice dilemma to defend abortion in a deontological context. So, many sacrifice dilemmas are really best suited to investigate our moral intuitions on concepts like DDE and the distinction between doing and allowing harm, while being less suited to the orthogonal issue of utilitarianism vs. deontology.
Returning to Kahane, in many cases the people in these trolley studies who choose the sacrifice options are not utilitarians in any plausible sense:
Utilitarians reject many conventional moral rules. But this rejection is certainly not the core of a utilitarian perspective. Its core is the impartial maximization of the good of all. The rejection of various deontological rules is just a consequence of that radical moral goal. In fact, the rejection of conventional moral rules is a feature utilitarianism shares with other views that may otherwise be diametrically opposed to it—such as egoism, which is likely to be the normative view that dominates the thinking of psychopaths…
But do we have any reason whatsoever to think that subjects who tend to make supposedly “utilitarian” judgments in sacrificial dilemmas view morality in more impartial terms compared to others? Not really. It is not only psychopaths and vmPFC patients who are more willing to endorse a “utilitarian” act when it also involves an element of self-interest, but also ordinary folk (Moore et al., 2008). And rates of “utilitarian” judgments are strongly influenced by whether they involve sacrificing (or saving) foreigners versus compatriots (Swann, Gómez, Dovidio, Hart, & Jetten, 2010), or strangers versus family members (Petrinovich, O’Neill, & Jorgensen, 1993)—let alone animals versus humans (Petrinovich et al., 1993).
Kahane and some colleagues directly investigate this issue in the 2015 paper ‘Utilitarian’ judgments in sacrificial moral dilemmas do not reflect impartial concern for the greater good:
Overall, endorsement of ‘utilitarian’ solutions to personal moral dilemmas was associated with lower wrongness ratings of the ‘utilitarian’ action (r = −.68, p < .001). Endorsement of ‘utilitarian’ solutions was associated with primary psychopathy (r = .29, p < .001) and marginally with reduced empathic concern (r = −.14, p = .06). Lower wrongness ratings of the ‘utilitarian’ action were associated with primary psychopathy (r = −.32, p < .001) and increased wrongness ratings with empathic concern (r = .17, p = .02). A multiple regression analysis testing the effects of psychopathy and empathic concern on wrongness judgments revealed that the two factors explained 10% of the variance in perceived wrongness of the utilitarian action (R2 = .10, F (2, 193) = 10.61, p < .001), but this effect was driven solely by primary psychopathy (β = −.1.11, p < .001).
…
In line with recent studies, we found that ‘utilitarian’ judgment was positively correlated with primary psychopathy and reduced empathic concern—traits that one would not expect to be associated with a genuine concern for the greater good. A regression analysis suggested that it was primary psychopathy rather than reduced empathic concern per se that drove the association with ‘utilitarian’ judgment.
Importantly, ‘utilitarian’ judgment was associated with more lenient assessment of immoral behavior in the Business Ethics measure. This association is directly between ‘utilitarian’ judgment and an amoral pattern of judgment, rather than, as in prior studies, only between ‘utilitarian’ judgments and reduced empathic concern or measures of antisocial personality traits. Notice, moreover, that this association was not fully explained by the correlation between ‘utilitarian’ judgment and psychopathy.
These results strongly suggest that so-called ‘utilitarian’ judgment is at least partly driven by a general antisocial or immoral tendency, rather than by a focused willingness to harm individuals in specific moral contexts.
So, social scientists who want to use sacrifice dilemmas to measure utilitarian attitudes need to be cautious in order to avoid merely measuring psychopathy. To that end Kahane and his colleagues propose a measurement called the Oxford Utilitarianism Scale that combine judgments in sacrifice scenarios and measurements of altruism in a joint measure of utilitarian attitude.
An Aside: Utilitarianism is Actually Altruistic
That experiments in sacrifice dilemmas tend to conflate utilitarian judgment with psychopathic/egoist judgment is in fact a problem for the sacrifice dilemma and not an indictment of the character of utilitarians. Ethical universalism (which I will refer to from now simply as “altruism”) has always been an essential component of the utilitarian ethic. This features directly in Jeremy Bentham’s “fundamental axiom”:
It is the greatest happiness of the greatest number that is the measure of right and wrong.
In his Utilitarianism, John Stuart Mill repeatedly emphasizes the altruistic nature of the principle of utility. He argues that the egoist takes no pleasure in the future of humanity:
To those who have neither public nor private affections, the excitements of life are much curtailed, and in any case dwindle in value as the time approaches when all selfish interests must be terminated by death: while those who leave after them objects of personal affection, and especially those who have also cultivated a fellow-feeling with the collective interests of mankind, retain as lively an interest in life on the eve of death as in the vigour of youth and health.
He argues that altruism is a powerful innate feeling, available to all but a select few, which is the rich soil from which the principle of utility grows:
Neither is it necessary to the feeling which constitutes the binding force of the utilitarian morality on those who recognize it, to wait for those social influences which would make its obligation felt by mankind at large. In the comparatively early state of human advancement in which we now live, a person cannot indeed feel that entireness of sympathy with all others, which would make any real discordance in the general direction of their conduct in life impossible; but already a person in whom the social feeling is at all developed, cannot bring himself to think of the rest of his fellow creatures as struggling rivals with him for the means of happiness, whom he must desire to see defeated in their object in order that he may succeed in his. The deeply-rooted conception which every individual even now has of himself as a social being, tends to make him feel it one of his natural wants that there should be harmony between his feelings and aims and those of his fellow creatures. If differences of opinion and of mental culture make it impossible for him to share many of their actual feelings perhaps make him denounce and defy those feelings- he still needs to be conscious that his real aim and theirs do not conflict; that he is not opposing himself to what they really wish for, namely, their own good, but is, on the contrary, promoting it. This feeling in most individuals is much inferior in strength to their selfish feelings, and is often wanting altogether. But to those who have it, it possesses all the characters of a natural feeling. It does not present itself to their minds as a superstition of education, or a law despotically imposed by the power of society, but as an attribute which it would not be well for them to be without. This conviction is the ultimate sanction of the greatest-happiness morality. This it is which makes any mind, of well-developed feelings, work with, and not against, the outward motives to care for others, afforded by what I have called the external sanctions; and when those sanctions are wanting, or act in an opposite direction, constitutes in itself a powerful internal binding force, in proportion to the sensitiveness and thoughtfulness of the character; since few but those whose mind is a moral blank, could bear to lay out their course of life on the plan of paying no regard to others except so far as their own private interest compels.
He argues that rival ethical theories (such as Kant’s deontology) inadvertently rediscover ethical universalism and the principle of utility:
It is no objection against this doctrine to say, that when we feel our sentiment of justice outraged, we are not thinking of society at large, or of any collective interest, but only of the individual case. It is common enough certainly, though the reverse of commendable, to feel resentment merely because we have suffered pain; but a person whose resentment is really a moral feeling, that is, who considers whether an act is blameable before he allows himself to resent it—such a person, though he may not say expressly to himself that he is standing up for the interest of society, certainly does feel that he is asserting a rule which is for the benefit of others as well as for his own. If he is not feeling this—if he is regarding the act solely as it affects him individually—he is not consciously just; he is not concerning himself about the justice of his actions. This is admitted even by antiutilitarian moralists. When Kant (as before remarked) propounds as the fundamental principle of morals, 'So act, that thy rule of conduct might be adopted as a law by all rational beings,' he virtually acknowledges that the interest of mankind collectively, or at least of mankind indiscriminately, must be in the mind of the agent when conscientiously deciding on the morality of the act. Otherwise he uses words without a meaning: for, that a rule even of utter selfishness could not possibly be adopted by all rational beings—that there is any insuperable obstacle in the nature of things to its adoption —cannot be even plausibly maintained. To give any meaning to Kant's principle, the sense put upon it must be, that we ought to shape our conduct by a rule which all rational beings might adopt with benefit to their collective interest.
Henry Sidgwick in The Method of Ethics offers this by way of rejecting egoism:
If the Egoist strictly confines himself to stating his conviction that he ought to take his own happiness or pleasure as his ultimate end, there seems no opening for any line of reasoning to lead him to Universalistic Hedonism [i.e. altruism] as a first principle; it cannot be proved that the difference between his own happiness and another’s happiness is not for him all-important. In this case all that the Utilitarian can do is to effect as far as possible a reconciliation between the two principles, by expounding to the Egoist the sanctions of rules deduced from the Universalistic principle,—i.e. by pointing out the pleasures and pains that may be expected to accrue to the Egoist himself from the observation and violation respectively of such rules. It is obvious that such an exposition has no tendency to make him accept the greatest happiness of the greatest number as his ultimate end; but only as a means to the end of his own happiness. It is therefore totally different from a proof (as above explained) of Universalistic Hedonism. When, however, the Egoist puts forward, implicitly or explicitly, the proposition that his happiness or pleasure is Good, not only for him but from the point of view of the Universe,—as (e.g.) by saying that ‘nature designed him to seek his own happiness,’—it then becomes relevant to point out to him that his happiness cannot be a more important part of Good, taken universally, than the equal happiness of any other person. And thus, starting with his own principle, he may be brought to accept Universal happiness or pleasure as that which is absolutely and without qualification Good or Desirable: as an end, therefore, to which the action of a reasonable agent as such ought to be directed.
Similarly, Derek Parfit asserts in a series of complex arguments in Part I of Reasons and Persons, that “self-interest theories” in ethics are collectively self-defeating due, ultimately, to their lack of impersonality and altruism.
A Warning For Rationalists
In my opinion, it’s good epistemic hygiene overall to have people ask each other interesting questions that bypass taboos, and to expect to hear actual answers in the form of bullet-biting, bets, forecasts, and other behaviors that promote taking ideas seriously without going insane. I’m not sure though that doing polls on X on sacrifice dilemmas plausibly figures into this hygiene, due to the Kahanian worry that this just ends up selecting for psychopathy. Without naming specific examples and people, I’m sure we can all think of case studies where rationalist social technologies ended up accidentally selecting for psychopaths. This dovetails with a hunch I’ve had that, for example, things like AskHole don’t really push people into thinking about uncomfortable topics more deeply, but rather are merely facilitating assortative mating along the axis of how dark triadic people are.
Is it actually possible to have productive discussions about utilitarianism? In my experience, they usually end up like:
A: I think it should be morally acceptable to do X.
B: It isn't okay to do X.
And then it just ends or becomes circular.
I don't think the polls on X select for psychopathy. It's showing what everyone thinks, but can't publicly say.