JOHANN FRICK

Contractualism and
Social Risk

I. INTRODUCTION

Contemporary nonconsequentialism is a family of views united less by a
positive doctrine than by skepticism toward central tenets of
consequentialist ethical thought. One such tenet, which is embraced by
most consequentialists but opposed by many nonconsequentialists, is
the notion of interpersonal aggregation.1 Ethical theories, like classical
utilitarianism, that defend interpersonal aggregation hold that in evaluating an action, we should sum the benefits and losses it imposes on
different people to obtain an aggregate quantity; this represents the
overall goodness of the action’s consequences. The rightness or wrongness of the action depends not on how it affects each individual, but on
the net balance of benefits over losses.
Aggregative reasoning of this kind often yields counterintuitive implications, especially in cases where it enjoins us to let a few people suffer

In writing this article I have accumulated many debts. For their very helpful comments
on earlier drafts, I thank Arthur Applbaum, Ralf Bader, Eric Beerbohm, Selim Berker, Dan
Brock, Norman Daniels, Tom Dougherty, Nir Eyal, Marc Fleurbaey, Mira Frick, Daniel
Halliday, Louis-Philippe Hodgson, Aaron James, Stephen John, Frances Kamm, Gregory
Keating, Paul Kelleher, Seth Lazar, Marc Lipsitch, Jeff McMahan, Andreas Mogensen, Ekédi
Mpondo-Dika, David O’Brien, Laura Ong, Derek Parfit, David Plunkett, Tim Scanlon, Lucas
Stanczyk, Victor Tadros, Gerard Vong, Dov Waisman, Alec Walen, and Daniel Wikler, as
well as two anonymous referees for Philosophy & Public Affairs. I would also like to thank
audiences and participants at the Moral and Political Philosophy Workshop at Harvard
University, the New York Early Career Ethics Workshop, the Lawrence S. Rockefeller
Seminar at Princeton, the Moral Philosophy Seminar at the University of Oxford, and the
Workshop on the Ethics of Social Risk at the Centre de recherche en éthique, Montreal.
1. For a dissenting voice in the nonconsequentialist camp, see S. Matthew Liao, “Who Is
Afraid of Numbers?” Utilitas 20 (2008): 447–61.
© 2015 Wiley Periodicals, Inc. Philosophy & Public Affairs 43, no. 3

176

Philosophy & Public Affairs

large losses so that many people can enjoy smaller benefits. A wellknown illustration is T. M. Scanlon’s
Transmitter Room Case: Jones has suffered an accident in a TV broadcasting station and is receiving extremely painful electrical shocks. If
we turn off the power to save him, billions of viewers will miss the last
half hour of the World Cup final.2
Many of us share Scanlon’s intuition that it would be wrong not to save
Jones from his agony, regardless of how many people are watching the
game. The benefit of watching a soccer match is trivial compared to the
agony of suffering strong electrical shocks. No matter how large the sum
of these benefits, it would be wrong to keep the power on. Defenders of
interpersonal aggregation, by contrast, seem committed to the view that
at some point the combined benefits to the viewers must become large
enough to morally outweigh Jones’s agony.
A common nonconsequentialist diagnosis for the inadequacy of
aggregative views is that they do not respect what John Rawls has called
“the separateness of persons.” Collections of people are not superindividuals. Unlike a single individual, who may rationally choose to
make some sacrifice in order to receive a stream of benefits, a group of
people lacks the requisite unity such that imposing a significant harm on
one person could straightforwardly be offset by conferring sufficient
benefits on others.
Seeking to formulate an account that better respects the separateness
of persons, leading nonconsequentialists have put forward so-called
competing claims models of moral rightness, according to which morality requires us to identify, by a series of pairwise comparisons, the action
or policy that satisfies the strongest individual claim or—its flipside—
generates the weakest individual complaint.3
The fullest development of this antiaggregative approach is the
contractualist moral theory defended by T. M. Scanlon. According to
Scanlon’s contractualist formula, an action is morally right if and only if
2. Adapted from T. M. Scanlon, What We Owe to Each Other (Cambridge, Mass.:
Harvard University Press, 1998), p. 235.
3. See, for instance, T. Nagel, “Equality,” in Mortal Questions (Cambridge: Cambridge
University Press, 1979), pp. 106–27; and T. Nagel, Equality and Partiality (Oxford: Oxford
University Press, 1991); see also T. M. Scanlon, What We Owe to Each Other, chap. 5.

177

Contractualism and Social Risk

it is justifiable to each person, that is, if a principle licensing the action
could not be reasonably rejected by any single individual for personal
reasons. Personal reasons, for Scanlon, are “reasons that are . . . tied to
the well-being, claims, or status of individuals in [a] particular position.”4
They exclude appeals to how an action will affect other individuals, as
well as to impersonal concerns, such as the goodness of the overall
outcome that the action will produce. Scanlon’s contractualist formula
thus rules out interpersonal aggregation in two distinct senses: On the
one hand, it excludes the “axiological” aggregation characteristic of classical utilitarianism. That is, it rejects the thought that the rightness of an
action is a function of the overall goodness of its consequences, which in
turn is determined by summing its good and bad effects for different
individuals. On the other hand, the contractualist formula also rules out
aggregation of a “normative” kind, in that it looks only at the personal
reasons of individuals and does not allow for the pooling of moral claims
or complaints across different individuals.
In addition, Scanlon affirms the following:
Greater Burden Principle: “It would be unreasonable . . . to reject a
principle because it imposed a burden on you when every alternative
principle would impose much greater burdens on others.”5
Scanlonian contractualism yields the intuitively correct answer in Transmitter Room. Turning off the power is the only course of action that is
justifiable to everyone, since Jones’s personal reasons for rejecting a
principle that allowed us to continue the transmission are much
weightier than any individual’s reasons against interrupting the broadcast. The fact that there are a large number of people with personal
reasons for not wanting the broadcast to be interrupted is deemed

4. Scanlon, What We Owe to Each Other, p. 219. For Scanlon, the personal reasons
invoked in reasonably rejecting a principle must be what he calls generic reasons. Generic
reasons are based not on the idiosyncratic interests and features of actual individuals,
but on what persons occupying this particular position would have reason to want “in
virtue of their situation, characterized in general terms.” Scanlon, What We Owe to Each
Other, p. 204.
5. T. M. Scanlon, “Contractualism and Utilitarianism,” in Utilitarianism and Beyond,
ed. Amartya Sen and Bernard Williams (Cambridge: Cambridge University Press, 1982), p.
111.

178

Philosophy & Public Affairs

morally irrelevant, since contractualism rejects the aggregation of individual claims.6
Cases like Transmitter Room, however, are quite rare. As Barbara
Fried remarks, outside the context of crime and war, most instances of
serious harm do not occur in situations where someone trades off certain
harms to known individuals against certain benefits to other known
individuals.7 (Let us call such situations interpersonal trade-offs under
certainty.) Rather, most harm to others results from risks that some agent
imposed on them, or failed to eliminate, in pursuit of some benefit for
them, for the agent, or for others. The question I wish to raise in this
article is whether contractualists can provide a plausible account of
such risky actions and omissions, while honoring their antiaggregative
commitments.
Most of my article will focus on one class of risky actions and omissions that, prima facie, are particularly troubling to the contractualist.
These actions are characterized by the following four features:
(1) The risky action or omission will affect a large number of individuals. Because of this, it is virtually certain that some people
will end up being burdened by it.
(2) The individual losses to those who are burdened (relative to the
baseline of some available alternative) are considerably greater
than the individual gains for those who are benefited.
(3) The action-type in question is rare, or rarely affects the same
people twice; as a result, we cannot assume that over time almost

6. Scanlon does allow that the numbers may count when we must choose between
preventing a smaller or a larger number of people from suffering an equivalent burden (for
example, saving one from death vs. saving two from death). But even here his reasoning
remains nonaggregative. In a choice between saving A or saving both B and C, it would be
wrong to flip a coin—but not because two deaths represent a greater total loss, or because
B and C can pool their claims to form a stronger collective claim. Rather, following Frances
Kamm, Scanlon argues that a principle which did not allow numbers to “break the tie” in
this case could be reasonably rejected by both B and C for failing to take account of the
value of saving their life, “since it permits the agent to decide what to do in the very same
way that it would have permitted had he not been present at all, and there was only one
person in each group.” Scanlon, What We Owe to Each Other, p. 232, following F. Kamm,
Morality, Mortality, vol. 2 (Oxford: Oxford University Press, 1993), pp. 116–17.
7. B. Fried, “Can Contractualism Save Us from Aggregation?” Journal of Ethics 16 (2012):
39–66.

179

Contractualism and Social Risk

everyone will benefit from a principle that permits actions of this
type to be performed.8
(4) The risky action or omission is intuitively permissible.
Many actions in public policy, particularly in the domains of public
health and of risk management, share these four features. (I will provide
a range of stylized examples over the course of this article.) The challenge
for the contractualist is to explain, without resorting to interpersonal
aggregation, how such risky actions could be permissible, given features
(1), (2), and (3). For want of an established label, I shall dub this the
problem of social risk.
As we shall see, cases of social risk force the contractualist to confront
a question about his theory that, I argue, has not received a fully satisfactory answer to date: Are an individual’s personal reasons for rejecting
a principle permitting the imposition of a given social risk a function of
the prospect that the risky action gives to each person ex ante, allowing us
to discount both benefits and harms by their improbability of occurring?
Or is justifiability to each person a function of the action’s outcomes ex
post? In that case, what would matter morally is not the ex ante likelihood of any given individual’s being benefited or harmed by the action,
but rather the near certainty that, ex post, some persons will turn out to
have been harmed by the action while others will have been benefited.
The contractualist faces an apparent dilemma: if he rejects the ex ante
view of justification, as Scanlon explicitly does in What We Owe to Each
Other, and appeals instead to an ex post view, as other authors have
urged,9 he seems committed to the intuitively unappealing conclusion
that most instances of social risk, as defined above, are unjustifiable. On
8. This feature blocks an appeal to what Scanlon calls “intra-personal aggregation,”
that is, “aggregation within each person’s life, summing up all the ways in which a principle
[prohibiting that kind of risky action] would constrain that life, rather than aggregation
across lives, adding up the costs or benefits to different individuals.” Scanlon, What We Owe
to Each Other, p. 237. Intrapersonal aggregation may allow the contractualist to justify the
performance of routine risky actions. While, on any given occasion, such a risky action may
impose significant individual losses on some while providing only relatively minor benefits
to others, over time almost everyone will come out ahead under a principle that permits
risky actions of this type to be performed. The kinds of risky actions that I am concerned
with in this article are not routine in this sense.
9. See, in particular, Sophia Reibetanz Moreau, “Contractualism and Aggregation,”
Ethics 108 (1998): 296–311. Elizabeth Ashford, “The Demandingness of Scanlon’s
Contractualism,” Ethics 113 (2003): 273–302, also assumes an ex post view of justification.

180

Philosophy & Public Affairs

the other hand, if he embraces an ex ante view, Scanlon fears that this
will move his contractualist theory too close to those aggregative views
that he wished to escape from in the first place.10
In this article, I argue that this dilemma can be overcome. I defend a
version of ex ante contractualism that gives a satisfactory response to the
problem of social risk, while also avoiding the excesses of aggregative
moral theories. My article proceeds as follows: To get a better grip on the
problem of social risk, Section 2 begins by introducing a stylized case of
social risk, and points out similarities with the more familiar problem of
interpersonal trade-offs under certainty. Section 3 presents a simple yet
seductive argument—the “Argument from Irrelevant Information”—
that supports an ex post view of contractualist justification. According to
this argument, the problem of social risk can be morally assimilated to
that of interpersonal trade-offs under certainty. Section 4 challenges this
view, arguing instead for an ex ante contractualism according to which
the problem of social risk is more akin to one of intrapersonal risk taking.
Section 5 returns to the Argument from Irrelevant Information, distinguishing three ways of fleshing it out. I argue that ultimately we must
reject the Argument from Irrelevant Information on any of these interpretations, and with it the ex post contractualism that it supports. In
Section 6, I critically examine T. M. Scanlon’s reasons for eschewing ex
ante contractualism, by considering what I call the problem of ex ante
rules. An ex ante rule is any rule the adoption of which, at some time t1, is
in everyone’s individual interest, but which licenses or requires some
agent to act at a later time t2 in a way that benefits some but significantly
burdens others. As we shall see, the worry that ex ante contractualism
may support morally indefensible ex ante rules is at the heart of
Scanlon’s objections to this view. Moreover, I show that existing
attempts by ex ante contractualists, such as Rahul Kumar, to overcome
10. This, at least, was Scanlon’s view in What We Owe to Each Other. In a more recent
article, “Reply to Zofia Stemplowska,” Journal of Moral Philosophy 10 (2013): 508–14,
Scanlon writes that his earlier opposition to ex ante contractualism was “a mistake,” and
credits an earlier draft of my article for changing his mind.
That ex ante contractualism cannot “save us” from aggregation is also the conclusion of
Barbara Fried, “Can Contractualism Save Us from Aggregation?” Two more optimistic
assessments of the prospects for ex ante contractualism, which can be read as complementing the case I make in this article, are given by Aaron James in “Contractualism’s (Not
So) Slippery Slope,” Legal Theory 18 (2012): 263–92; and Rahul Kumar, “Risking and Wronging,” Philosophy & Public Affairs 43 (2015): 27–49.

181

Contractualism and Social Risk

this difficulty are not fully satisfactory.11 However, by adopting what I call
the Decomposition Test, according to which an action is justifiable
if and only if the actions it licenses us to perform are justifiable to each
person at each temporal stage, I believe that Scanlon’s misgivings can be
overcome.
Unfortunately, ex ante contractualism’s success at dealing with the
problem of social risk comes at a theoretical price, which hitherto has
received scant attention from defenders of this view. In Section 7, I draw
out some of the implications that ex ante contractualism has for a related
problem: the question of “identified” vs. “statistical” lives. I argue that,
contrary to received wisdom, ex ante contractualism can lend a measure
of normative support to our common psychological propensity to give
greater weight to the saving of “identified” over “statistical” lives. The
problem, as I show in Section 8, is that, under certain circumstances, ex
ante contractualism would lead us to privilege the saving of one identified life over any number of statistical lives.12 The most promising way of
avoiding this extreme conclusion, I argue in Section 9, is by scaling back
the ambitions of contractualism as a moral theory. Instead of providing
a complete account of what it is for actions to be right or wrong, the
contractualist competing claims model is better understood as capturing
an important class of pro tanto moral reasons that contribute to making
actions right or wrong, but that do not by themselves determine an
action’s rightness all things considered.
II. SOCIAL RISK AND INTERPERSONAL TRADE-OFFS UNDER CERTAINTY

Consider the following stylized case of social risk:
Mass Vaccination (Unknown Victims): One million young children are
threatened by a terrible virus, which is certain to kill all of them if we
do nothing. We must choose between mass producing one of two
vaccines (capacity constraints prevent us from producing both):
• Vaccine 1 is certain to save every child’s life. However, the vaccine
will not provide complete protection against the virus. If a child
11. Kumar, “Risking and Wronging.”
12. James, “Contractualism’s (Not So) Slippery Slope,” and Kumar, “Risking and
Wronging,” the two main proponents of ex ante contractualism in the literature, both fail
to discuss this problem.

182

Philosophy & Public Affairs

receives Vaccine 1, the virus is certain to paralyze one of the child’s
legs, so that he or she will walk on crutches for the rest of his or her
life.
• Vaccine 2 is risky. It gives every child a 999/1000 chance of surviving the virus completely unharmed. However, for every child there
is a 1/1000 chance that Vaccine 2 will be completely ineffective and
that the child will be killed by the virus. (Assume that the outcomes
for different children are probabilistically independent.) Call the
children who end up dying the luckless children.
Whichever of the two vaccines we choose to produce will be administered to all one million children.13
Before I go on, a brief note about the notions of “chance” and “probability,” as I employ them in this and subsequent cases. Notwithstanding the posits of contemporary quantum physics, I shall assume that
natural processes at the macro-physical level that we deal with in medicine and most other domains of risky human activity are, for all intents
and purposes, deterministic. When using the terms “probability” or
“chance” in this and all following examples, I shall therefore assume
that we are speaking not about objective indeterminacy at the level of
physical reality itself, but about epistemic probability, which reflects
our incomplete knowledge of the state of the world and the laws of
nature. The notion of epistemic probability that I will be employing is
evidence-relative: To say that, from the perspective of an agent, some
event e has an epistemic probability p of occurring is to say that, given
the evidence available to the agent, her rational degree of credence in
the proposition “e will occur” would be p. (To keep things simple, I
shall assume that the decision makers’ actual degrees of credence track
what the available evidence makes it rational for them to believe.) In
Section 5.C, I revisit this assumption and argue that, for the purposes of
morally evaluating risky actions, it typically makes no difference
whether the risk in question corresponded to objective or to “merely”
epistemic chances.
13. I specify that the patients are young children in order to bracket the complicating
factor of patient autonomy; an adult patient would presumably have the right to refuse
either treatment.

183

Contractualism and Social Risk

How ought a contractualist to go about deciding how to act in Mass
Vaccination (Unknown Victims)? One way of thinking about this case is
by analogy with the following interpersonal trade-off under certainty:
Mass Vaccination (Known Victims): The threat situation is as in Mass
Vaccination (Unknown Victims) above. This time, we must choose
between mass producing one of the following two vaccines:
• Vaccine 1, as above, is sure to save every child’s life, at the cost of
paralyzing one of their legs.
• Vaccine 3 is sure to allow 999,000 children to survive the virus
completely unharmed. However, because of a known particularity
in their genotype, Vaccine 3 is certain to be completely ineffective
for 1,000 identified children. These doomed children are sure to be
killed by the virus if we choose Vaccine 3.
From what was said in the previous section, it is clear how a
Scanlonian contractualist ought to think about Mass Vaccination
(Known Victims). The individual burden of becoming paralyzed in one
leg, though significant, is not even close to that of losing one’s life at a
young age. Since contractualism does not allow the aggregation of
claims, the Greater Burden Principle implies that no one could reasonably reject a principle requiring us to produce Vaccine 1, whereas a principle licensing us to choose Vaccine 3 could be reasonably rejected by
each of the 1,000 doomed children.
Now return to Mass Vaccination (Unknown Victims). Note that, due
to the Law of Large Numbers, we can predict with a high degree of
confidence that the overall pattern of outcomes from administering
Vaccine 2 would look a lot like that of administering Vaccine 3 in my
second case. Although no particular child is certain to die, it is a statistical certainty that some children will be killed by the virus if we administer Vaccine 2, and the likelihood that roughly 1,000 (±100) will die is
greater than 99 percent.14
999 ⎞ 1 ,000,000
14. The probability that at least one child will be killed by Vaccine 2 is 1 − ⎛
,
⎝ 1,000 ⎠
which is very close to 1. By Chebyshev’s Inequality, the likelihood that roughly 1,000 children (±100) will be killed is > 0.99. Given that contractualism prohibits aggregation, only the
former number really matters to the argument.

184

Philosophy & Public Affairs

Moreover, the harm of death is just as great for a luckless child as for
a doomed child. Holding constant the fact of actually suffering a harm, it
is typically no better for a person that she had some chance of escaping
the harm rather than being certain to suffer it. (I return to this point in
Section 5.)
III. THE ARGUMENT FROM IRRELEVANT INFORMATION AND
EX POST CONTRACTUALISM

Given these two parallels—a near-identical overall pattern of outcomes
and equivalent individual harms—a number of nonconsequentialist
philosophers have maintained that cases of social risk can, under certain
conditions, be morally assimilated to interpersonal trade-offs under certainty. In the following, I will concentrate on an argument by Sophia
Reibetanz Moreau.15 Elsewhere,16 I have contended with similar arguments by Marc Fleurbaey and Alex Voorhoeve.17
All of these philosophers deploy versions of the following basic argument, which, applied to my two examples, runs as follows:
The Argument from Irrelevant Information: In Mass Vaccination
(Known Victims), selecting Vaccine 3 is unjustifiable to the doomed
children, since this would impose on them a greater individual
burden than Vaccine 1 imposes on anyone. But in Mass Vaccination
(Unknown Victims), we know that, if we choose Vaccine 2, some children will end up in the same position that the doomed children occupied in Mass Vaccination (Known Victims). They, too, will lose their
lives, whereas under Vaccine 1 the greatest harm that any child would
have had to suffer is a paralyzed leg. To be sure, we do not yet know
who these luckless children are. But surely, the identity of the eventual
victims is irrelevant information. Whoever will lose their life through
Vaccine 2 will have as strong a complaint against the decision maker
15. Reibetanz Moreau, “Contractualism and Aggregation.”
16. Johann Frick, “Uncertainty and Justifiability to Each Person: Response to Fleurbaey
and Voorhoeve,” in Inequalities in Health: Concepts, Measures, and Ethics, ed. Nir Eyal,
Samia Hurst, Ole Norheim, and Daniel Wikler (New York: Oxford University Press, 2013).
17. Alex Voorhoeve and Marc Fleurbaey, “Decide as You Would with Full Information!
An Argument against Ex Ante Pareto,” in Eyal, Hurst, Norheim, and Wikler, Inequalities in
Health.

185

Contractualism and Social Risk

as a doomed child has against Vaccine 3. Therefore, Vaccine 1 is the
only option that is justifiable to each person in either case.
For proponents of the Argument from Irrelevant Information, the
strength of an individual’s personal reasons for rejecting a risky action
depends not on how likely they themselves were to suffer a burden, but
on how likely it was that someone would. Thus, Sophia Moreau writes:
As long as we know that acceptance of a principle will affect someone
in a certain way, we should assign that person a complaint that is
based upon the full magnitude of the harm or benefit, even if we
cannot identify the person in advance. It is only if we do not know
whether acceptance of a principle will affect anyone in a certain way
that we should allocate each individual a complaint based upon his
expected harms and benefits under that principle.18
Moreau subscribes to what is called an ex post view of justification,
according to which the relative strength of an individual’s harm-based
complaint turns not on the ex ante prospect it offered her individually,
but on the foreseeable distribution of outcomes across individuals that
the action will produce ex post. Paired with a ban on the aggregation of
claims, this ex post view strongly privileges the perspective of the individual who turns out to be most burdened under a given principle. It is to
her that justification must be addressed, typically by showing that any
alternative principle would have ended up imposing an even greater
burden on her or on someone else.
Although the Argument from Irrelevant Information is seductive, I
believe that ex post contractualism is hard to accept. Note, in particular,
that it would dramatically contradict many of our ordinary moral convictions: in real life, we often impose social risks that closely resemble
that of choosing Vaccine 2 in Mass Vaccination (Unknown Victims).
Thus, it is commonly deemed morally unproblematic to systematically
inoculate young children against certain serious but nonfatal childhood
diseases where there is a remote chance of fatal side effects from the
inoculation itself. But this is not because it is unlikely that inoculation
will ever lead to disaster for some unlucky children. Given the large
18. Reibetanz Moreau, “Contractualism and Aggregation,” p. 304.

186

Philosophy & Public Affairs

number of children inoculated each year, it is a statistical certainty that
some small number of them will develop fatal complications. If ex post
contractualism were correct, a policy of systematic inoculation would be
very hard to defend. Indeed, it would be as hard to defend as a policy that
sacrificed the lives of some small known group of children each year in
order to save the others from contracting nonfatal childhood diseases.

IV. THE ARGUMENT FROM THE SINGLE-PERSON CASE AND
EX ANTE CONTRACTUALISM

Given these problems with ex post contractualism, it is worth exploring
a different approach. A natural thought is that, instead of assimilating
Mass Vaccination (Unknown Victims) to an interpersonal trade-off
under certainty, we could compare it to a single-person decision under
risk. Consider:
Individual Vaccination: A single young child (call her Clara) is threatened by a terrible virus, which is certain to kill her if we do nothing. As
Clara’s guardians, we must decide which of two available vaccines to
give her. As before,
• Vaccine 1 is certain to save Clara’s life, but at the cost of one of her
legs becoming paralyzed by the virus.
• Vaccine 2 will give Clara a 999/1000 chance of surviving the virus
completely unharmed; however, there is a 1/1000 chance that
Vaccine 2 will be completely ineffective for Clara and that she will
be killed by the virus.
How ought we to act? In von Neumann-Morgenstern’s expected utility
theory, the effect of a utility loss on a person’s overall expected utility is
weighted by the probability of its occurrence; this means that large but
unlikely losses can be equivalent in expected utility to smaller but more
likely losses (likewise for gains). Hence, according to expected utility
theory, the prospect offered to Clara by Vaccine 1 is worse than that offered
by Vaccine 2 if and only if her utility from becoming paralyzed and walking
on crutches for the rest of her life is less than the weighted sum of her
utility loss from death (weighted by a factor of 1/1000) plus the benefit of
a life in full health (weighted by a factor of 999/1000). Many of us would

187

Contractualism and Social Risk

agree that this is the case, and hence that it would be morally right for
Clara’s guardians to choose Vaccine 2 on her behalf.
This result is not an artifact of embracing expected utility theory.
Some versions of prioritarianism, for instance, maintain that as Clara’s
guardians we ought not necessarily select the treatment that gives her
the greatest expected utility ex ante. Rather, we should assign disproportionate normative weight to avoiding an outcome that is very bad for
Clara (in this case, her death).19 Thus, even if Clara’s expected utility from
receiving Vaccine 2 were somewhat greater than her expected utility
from Vaccine 1, a prioritarian might still opt to give her Vaccine 1, because
this is sure to avert the outcome that is worst for Clara. In the present
case, however, Clara’s expected utility from receiving Vaccine 2 is arguably considerably greater than her expected utility from Vaccine 1. Given
this, I claim, a reasonable prioritarian would concur that it would be
justifiable for Clara’s guardians to choose Vaccine 2 on her behalf. For a
young child, escaping certain paralysis in one leg for the rest of her life is
worth a 1/1000 risk of death.
Our response in Individual Vaccination points the way to a different
way of thinking about the imposition of social risks. I call this the Argument from the Single-Person Case:
Argument from the Single-Person Case: For contractualists, the rightness of an action is a function of each individual’s personal reasons for
rejecting a principle that licenses the action. An act is wrong if and only
if there is someone who can complain that we failed to treat her in a way
that was justifiable to her, not because its consequences were impersonally bad. In assessing the rightness of a risky action, we must focus
on how it affects each individual, compared to the possible alternatives. To better see how Mass Vaccination (Unknown Victims) affects
each person individually, this case might be analytically decomposed
into one million single-person gambles that occur in parallel. But
what do we see if we do this? The prospect offered to each child by
19. I set aside the question whether this version of prioritarianism, which applies the
view not just to interpersonal trade-offs but to intrapersonal gambles as well, is the most
plausible way of fleshing out the view. For discussion of this point, see Michael Otsuka and
Alex Voorhoeve, “Why It Matters That Some Are Worse Off Than Others: An Argument
against the Priority View,” Philosophy & Public Affairs 37 (2009): 171–99; and Derek Parfit,
“Another Defense of the Priority View,” Utilitas 24 (2012): 399–440.

188

Philosophy & Public Affairs

administering Vaccine 2 is exactly the same as the one given to Clara in
Individual Vaccination, and the alternative option is the same as well.
Hence, by mass producing and administering Vaccine 2, we do for each
child what a guardian, concerned solely with that child’s interests,
ought to have done for her in a single-person case. Given this, how
could any child in Mass Vaccination (Unknown Victims) reasonably
reject a principle that licenses us to choose Vaccine 2?20
Underpinning this argument is a different way of thinking about the
justification of social risk. According to ex ante contractualism, the
strength of someone’s personal reasons for rejecting a principle licensing a risky action depends on the quality of the prospect that the action
gave her ex ante. A person’s harm-based complaint against a loss she
suffers must, therefore, be discounted by her ex ante unlikelihood of
suffering a loss and by her ex ante likelihood of benefiting from the risky
action. If the person’s ex ante prospect from the risky action was good
enough, she may have no reasonable complaint under the Greater
Burden Principle, even if she is unlucky and ends up suffering a greater
loss than anyone would have suffered under an available alternative.21
20. Alternatively, consider the following sequential decomposition of Mass Vaccination
(Unknown Victims): A doctor must, over a period of time, see one million young children,
each in the same situation as Clara in Individual Vaccination. Due to the Law of Large
Numbers, the doctor can foresee that, if he gives each child Vaccine 2, at some point
misfortune will strike and some child will be killed by the virus. Does this mean that, while
it would be justifiable to give Clara Vaccine 2 in Individual Vaccination, it would be unjustifiable for the doctor to adopt a policy of giving every child Vaccine 2 in the sequential
case? This is what ex post contractualism implies, but it seems scarcely credible. The
potential benefits for each child of receiving Vaccine 2 in the sequential case are just as
great as they were for Clara in Individual Vaccination, and the risk faced in exchange is no
greater than what was deemed acceptable to Clara in that case. It seems implausible that a
policy of giving every child Vaccine 2 should be unjustifiable to a child, not because of how
it is likely to affect that child, but just because, under that policy, other children will, at
different points in time, be exposed to the same probabilistically independent risk. For an
argument along similar lines, see Tom Dougherty, “Aggregation, Beneficence, and
Chance,” Journal of Ethics and Social Philosophy 72 (2013): 1–19.
21. This is not to claim that prospects have nonderivative value or disvalue in themselves. As I will argue in Section 5, people are not made worse off by the mere risk of
suffering a loss, nor do they benefit from the mere chance of receiving a benefit, holding
everything else constant. But while what ultimately matters to individuals’ well-being is
whether they end up being benefited or harmed, what might matter morally is to ensure a
justifiable distribution of prospects. I owe this point to Derek Parfit, “Justifiability to Each
Person,” Ratio 16 (2003): 368–90.

189

Contractualism and Social Risk

Ex ante contractualism, as I am proposing it, must be sharply distinguished from the contractualist rule-utilitarian view espoused by John
Harsanyi.22 According to Harsanyi, in order to decide an interpersonal
trade-off under certainty, such as Mass Vaccination (Known Victims), in
a manner that is justifiable to each person, we should imagine all
affected parties placed behind a hypothetical veil of ignorance that
deprives them of all knowledge of their personal identity. The right
course of action is that which all parties would, in this situation, endorse
on grounds of self-interest.
To Harsanyi, the hypothetical veil of ignorance device offers two principal attractions: First, it promotes impartial and unbiased decision
making by modeling our moral principles on the choices of hypothetical
deliberators who lack the ability to tailor principles to their personal
traits and needs. Behind the veil of ignorance, even selfish deliberators
are forced to give fair consideration to the interests of every person
concerned—for all they know, they might turn out to be that person.
Second, the veil of ignorance device promises to reduce a difficult interpersonal trade-off to a more tractable problem of intrapersonal choice
under risk. Since none of the affected parties possesses information that
would give them reason to prefer a different course of action than
anyone else, we can expect deliberation behind the hypothetical veil of
ignorance to render a unanimous verdict.
In Mass Vaccination (Known Victims), this verdict would be to
produce Vaccine 3. Given no information about her identity, Harsanyi
maintains, each child must assume that there is a 999/1000 chance that
Vaccine 3 will work for her and only a 1/1000 chance that it will not (if she
turns out to be one of the doomed children).23 This intrapersonal gamble
is equivalent to that of Individual Vaccination, and there we concluded
that it would be in a child’s interest to take the risk. Therefore, according
to Harsanyi, choosing Vaccine 3 can be justified to each person and is
morally right.
22. See, for instance, John Harsanyi, “Morality and the Theory of Rational Behavior,” in
Sen and Williams, Utilitarianism and Beyond, pp. 39–62.
23. John Rawls’s more famous use of the veil of ignorance device differs from Harsanyi’s
in this regard. Whereas Harsanyi assumes that every deliberator behind the veil has an
equal epistemic probability of occupying any position beyond the veil, Rawls himself
eschews this equiprobability assumption in favor of a “thick” veil of ignorance, which
deprives the deliberators of all probabilistic knowledge.

190

Philosophy & Public Affairs

Scanlon, however, rejects Harsanyi’s hypothetical veil of ignorance
approach, arguing that it mischaracterizes the nature of justification to
individuals. By modeling interpersonal trade-offs on analogy with a
single individual’s choices under risk, the hypothetical veil of ignorance
mechanism fails to give each person the separate and individual concern
that she is due. That it is often rational to take risks in pursuit of greater
benefits in our own lives does not, ipso facto, justify making someone
suffer burdens for the good of other people. Thus, “the question of what
everyone could reasonably agree to or what no one could reasonably
reject” is “a quite different question [from] what would maximize the
expectations of a single self-interested person choosing in ignorance of
his true position.”24 For a moral principle to be valid is for it to be justifiable to each person from her own point of view, without artificial informational restrictions.
Unlike Harsanyi’s hypothetical veil of ignorance method, which
works by depriving people of information that they in fact have, ex ante
contractualism of the kind I am proposing comes into its own in
situations where our inability to foresee the individual outcomes of our
actions places us behind what might be called a natural veil of
ignorance. It is natural uncertainty itself, not the strictures of a hypothetical veil of ignorance, that forces people to deliberate in the
absence of complete knowledge about how they will be affected by the
action in question. To claim that such situations can be illuminatingly
analyzed by analogy with intrapersonal gambles under risk is not to
commit ourselves to a view that reduces interpersonal justification in
general to a question of self-interested choice under informational
restrictions.
Indeed, if ex ante contractualism is sound, it suggests a sharp distinction between the two Mass Vaccination cases. Choosing Vaccine 3 in
Mass Vaccination (Known Victims) means the doomed children face the
prospect of certain death. By contrast, administering Vaccine 2 in Mass
Vaccination (Unknown Victims) gives no child a comparably bad prospect. While it is certain that some children will die, there is no child who
is certain to die if we administer Vaccine 2. Indeed, given the significant
burden of partial paralysis under Vaccine 1, a course of action that

24. Reibetanz Moreau, “Contractualism and Utilitarianism,” p. 122.

191

Contractualism and Social Risk

promises to avoid this certain burden at the cost of a tiny risk of death
seems in every child’s interest. Given this, selecting Vaccine 2 is justifiable to every child, according to ex ante contractualism.

A. The Case of Knowable Victims
In discussing the contrast between Mass Vaccination (Known Victims)
and Mass Vaccination (Unknown Victims), I have so far focused on two
classes of cases: cases where we know the identities of individuals who
will suffer if we perform a certain action and cases where, given the
available evidence and current scientific know-how, we cannot know, at
the time of acting, who will suffer if we act in a certain way. But this
distinction is not exhaustive.
Consider, briefly, an intermediate case, which is like Mass Vaccination
(Known Victims), except that the children who will be killed by the virus
if we choose Vaccine 3 are not currently known but merely knowable, in
the sense that we have the necessary evidence and wherewithal to determine their identities prior to administering Vaccine 3. For instance,
suppose we know that Vaccine 3 will be completely ineffective for a child
if and only if she carries a rare gene G. Moreover, we know that there are
1,000 such children among the one million threatened by the virus. What
we do not currently know is which children carry G. However, a simple
and costless genetic test would allow us to find out.25 Call this case Mass
Vaccination (Knowable Victims).
Intuitively, the fact that the victims in this case are not known but
merely knowable does not make a moral difference. “Surely,” the
thought goes, “if vaccinating would be unjustifiable in Mass Vaccination
(Known Victims) because we already know the identities of the doomed
children, our ignorance of the doomed children’s identities in Mass Vaccination (Knowable Victims) would be a very poor excuse for choosing
Vaccine 3. For this lack of information about the doomed children’s
identities is one that we are entirely free to remove by carrying out the
simple genetic test.”
25. Alternatively, imagine that the test has already been carried out, and that its
results—though not yet known to us—are stored in a filing cabinet where we could easily
consult them.

192

Philosophy & Public Affairs

Ex ante contractualism supports this intuitive verdict.26 To see why,
return first to the contrast between Mass Vaccination (Unknown Victims)
and Mass Vaccination (Known Victims). According to the Argument from
the Single-Person Case, the crucial difference between these cases is that
in the former we can say to each child: “To the best of our knowledge,
choosing Vaccine 2 is highly likely to benefit you, and has only a tiny
chance of turning out to your disadvantage. Indeed, we could justify
giving you this vaccine even in a single-person case, where furthering your
interests was our sole concern.” By contrast, we could not say this to each
child in Mass Vaccination (Known Victims), in attempting to justify
Vaccine 3. For in that case, there are some individuals—the doomed
children—of whom we know that they will die if we choose Vaccine 3. For
a doomed child, choosing Vaccine 3 does not correspond to what we
ought to do in a case where her interests were our sole concern.
I maintain that in the respects highlighted by the Argument from the
Single-Person Case, Mass Vaccination (Knowable Victims) is equivalent
to Mass Vaccination (Known Victims). As in Mass Vaccination (Known
Victims), we could not justify Vaccine 3 by saying to each child: “To the
best of our knowledge, choosing Vaccine 3 is highly likely to benefit you,
and has only a tiny chance of turning out to your disadvantage. We could
justify giving you this vaccine even in a single-person case.” Both halves
of this statement are false.
Consider the first half first. Granted, it is true that
(A) In Mass Vaccination (Knowable Victims), given our present
ignorance about which children have gene G, there is no child of
whom we know that will die if we choose Vaccine 3.
Indeed,
(B) In Mass Vaccination (Knowable Victims), given our present
ignorance about which children have gene G, it is rational to
believe, of any given child, that it is highly likely to benefit if we
produce Vaccine 3, and very unlikely to be burdened.
Yet both (A) and (B) are things we can affirm only as long as we do not
avail ourselves of all the freely available evidence, namely, by carrying
26. I thank an anonymous referee for pressing me to clarify my reasoning about this
point.

193

Contractualism and Social Risk

out the simple and costless genetic test. However, something that we
can only affirm without taking into account freely available evidence is
not something we can affirm “to the best of our knowledge.” Hence,
neither (A) nor (B) can support
(C) In Mass Vaccination (Knowable Victims), it is true of every
child that, to the best of our knowledge, choosing Vaccine 3 is
highly likely to benefit the child, and has only a tiny chance of
turning out to her disadvantage.
Nor is it true that, by choosing Vaccine 3 in Mass Vaccination (Knowable Victims), we are treating each child in a way that would have been
justifiable in a single-person case. Consider a single-person analogue
involving only Clara and her doctor: The doctor can administer a genetic
test, which will determine whether Clara carries G or not. Given that this
test is simple and costless, and will help the doctor to select the appropriate treatment, there is no justification for failing to administer the test.
But if the test indicates that Clara carries gene G, it will not be justifiable
for the doctor to administer Vaccine 3. He must administer Vaccine 1.
Consequently, we cannot say to each child in Mass Vaccination (Knowable Victims) that by giving him Vaccine 3, we are treating him in a way
that would have been justifiable in a single-person case. This will be false
for those children who carry G.
I conclude that if ex ante contractualists are right to use the Argument
from the Single-Person Case to draw a moral line between cases like
Mass Vaccination (Unknown Victims) and Mass Vaccination (Known
Victims), then cases involving victims whose identities are not presently
known but are knowable should often be treated on analogy with cases
involving known victims—at least if it is simple and costless to ascertain
their identities.
Suppose, by contrast, that although we could find out who
carries gene G in Mass Vaccination (Knowable Victims), doing so
would be extremely costly for us. Intuitively, this seems to bring this case
closer to Mass Vaccination (Unknown Victims), where information
about the identities of the victims is simply unobtainable prior to
acting.
I believe that ex ante contractualism has the resources to underwrite
this intuition as well, though I can only sketch a tentative proposal. Once
again, it is helpful to first consider a single-person analogue of the case in

194

Philosophy & Public Affairs

question. Suppose that, in principle, the doctor has the ability to determine whether Clara carries the rare gene G, but that doing so would be
extremely costly to the public purse (Clara herself cannot bear these
costs). In this case, the doctor may have a valid justification for not
carrying out the genetic test, and for instead giving Clara the treatment
that gives her the best prospect, given ignorance about her exact genotype, namely, Vaccine 3. He might say to Clara: “To the best of my knowledge, given justifiable limits on the resources we can be expected to
expend in gathering further information about your particular case,
Vaccine 3 is highly likely to benefit you, and has only a tiny chance of
turning out to your disadvantage.”
This reasoning carries over to Mass Vaccination (Knowable Victims).
Unlike in the costless case we discussed above, choosing Vaccine 3 in the
costly case may be justifiable to every child. For in this case, by giving all
children Vaccine 3, we would be doing for them the same thing that
would have been justifiable to each in a single-person case.

V. REJECTING THE ARGUMENT FROM IRRELEVANT INFORMATION AND
EX POST CONTRACTUALISM

Since ex ante contractualism avoids some of the implausible implications of the ex post view and seems independently attractive, it pays to
take a closer look at the ex post contractualist’s Argument from Irrelevant Information, to see whether we can find fault with it.
Consider the following reconstruction of the Argument from Irrelevant Information:
(1) An action is permissible if and only if it is justifiable to each
person.
(2) In Mass Vaccination (Known Victims), selecting Vaccine 3 cannot
be justified to the doomed children.
(3) In Mass Vaccination (Unknown Victims), we know that, if we
select Vaccine 2, the luckless children will suffer the same losses
that the doomed children suffer under Vaccine 3, although we
cannot know the identities of the luckless children.
(4) The fact that we cannot know the identities of the luckless children is irrelevant to the question of whether selecting Vaccine 2 is
justifiable to each person.

195

Contractualism and Social Risk

(5) In Mass Vaccination (Unknown Victims), there are some children to whom selecting Vaccine 2 cannot be justified, although
we cannot know the identities of these children. [from (2), (3),
and (4)]
∴ Selecting Vaccine 2 is impermissible in Mass Vaccination
(Unknown Victims). [from (1) and (5)]
The crucial step in this argument is premise (4). Proponents of the
Argument from Irrelevant Information assume that the lack of information about individual outcomes that characterizes cases of social
risk is a merely epistemic phenomenon, devoid of moral significance. It
is morally irrelevant, they claim, that we cannot know who will be
harmed by the risky action, as long as we know that some persons will
be harmed. Instead of affecting whether Vaccine 2 is justifiable to each
person, our lack of information about individual outcomes merely
masks the identity of those children to whom the choice of Vaccine 2
cannot be justified.
Premise (4), however, directly contradicts what defenders of ex ante
contractualism affirm. And they, as we saw, have an argument for rejecting premise (4): given lack of information about individual outcomes, it
is in each patient’s ex ante interest to take the gamble that Vaccine 2
offers. By administering Vaccine 2, we thus do for each child what we
would do on her behalf in a single-person case, where that child’s wellbeing was our only concern. The same cannot be said in an interpersonal
trade-off under certainty.
Lest they beg the question against their opponent, defenders of ex
post contractualism thus need an independent argument for affirming
premise (4). Let me review what seem to me the three strongest
contenders.
A. The Argument from Certain Loss
The first and, I believe, the most natural way of reading the passage from
Moreau that I cited above is as proposing what I shall term the Argument
from Certain Loss. According to this argument, information about individual outcomes is not needed in cases of social risk, because whoever
ends up being burdened by the risky action will have a complaint that is
augmented by the fact that it was certain that someone would be burdened. Recall Moreau: “As long as we know that acceptance of a

196

Philosophy & Public Affairs

principle will affect someone in a certain way, we should assign that
person a complaint that is based upon the full magnitude of the harm or
benefit, even if we cannot identify the person in advance.”
The Argument from Certain Loss, however, implies an odd asymmetry
between single-person gambles and cases of social risk-imposition. A
proponent of the Argument from Certain Loss would agree that a principle licensing us to choose Vaccine 2 in Individual Vaccination is justifiable to Clara. Even if Clara is unlucky and the gamble turns out badly for
her, she has no weighty complaint. After all, in this single-person case, the
likelihood that someone will be killed is the same as the likelihood that
Clara will be killed. By contrast, the choice of Vaccine 2 is said to be
unjustifiable in Mass Vaccination (Unknown Victims), since here it is a
statistical certainty that some child will be killed. What, therefore, makes
the moral difference between the single-person and the social case is our
enhanced knowledge of the overall pattern of outcomes—not any difference in the attractiveness of the gamble offered to each individual.
But why, we may wonder, should a luckless child have a greater complaint than Clara, just because, even if he had not been unlucky, it is
statistically certain that some other child would have died instead? This
looks suspiciously like a new form of interpersonal aggregation: the combination of complaints by different individuals at different possible
worlds, depending on who happens to be unlucky at that possible world.
But if contractualists reject the combination of claims by different people
at the same possible world, then a fortiori they ought not to allow different
people to aggregate their complaints across different possible worlds.
B. The Argument from Lack of Concern
According to the next argument, the lack of information about individual
outcomes in Mass Vaccination (Unknown Victims) does not matter,
because we know that whoever ends up being burdened could make the
following complaint: given that we knew for certain that some children
would die if we chose Vaccine 2—an outcome that we could have
avoided by opting for Vaccine 1 instead (which would not have imposed
a similarly severe loss on anyone)—our willingness to nonetheless opt
for Vaccine 2 betrays a lack of concern for the eventual losers.
This argument also misses its mark: it might be claimed that our
action shows too little concern for avoiding that there will be losers. This,

197

Contractualism and Social Risk

however, expresses a preoccupation with the overall shape of the
outcome, as might arise from an impersonal moral principle such as telic
egalitarianism or the priority view. As I argued above, contractualism’s
exclusive focus on personal reasons does not allow us to appeal to such
impersonal principles.
By contrast, what no luckless child can claim is that the risky action
showed too little concern that he will be a loser. Again, our action takes
no greater risks with the well-being of any child than its analogue in
Individual Vaccination, where defenders of ex post contractualism
agreed that taking the gamble on Clara’s behalf did not show insufficient
concern for her. Unequal outcomes are not always evidence of unequal
concern or treatment, especially when they are produced by a chancy
causal process.
C. The Argument from Determinism
Let us consider one final argument for the ex post perspective. We are
conducting our discussion under the working assumption that at the
macro-physical level that we deal with in most cases of social riskimposition, nature is, for all intents and purposes, deterministic. All
probabilities we encounter in these contexts are merely epistemic, that is,
owed to our incomplete knowledge of the state of the world and the laws
of nature that govern it, not objective, that is, due to indeterminacy at the
level of physical reality itself.27
But if this assumption is true, might it not spell trouble for the ex ante
contractualist? After all, if determinism is true, then someone who loses
27. How exactly to characterize the notion of “objective” probability remains a hotly
debated topic in contemporary metaphysics. The intuitive contrast with the “epistemic”
notion of probability that I have been working with is, nonetheless, clear enough. Recapitulating from above: to say that, from the perspective of an agent, some event e has an
epistemic probability p of occurring is to say that, given the evidence available to the agent,
her rational degree of credence in the proposition “e will occur” would be p. Note that,
given incomplete knowledge of the world, p can be > 0 even for events that are, as a matter
of fact, physically impossible. By contrast, to claim that e has an objective probability p of
occurring is to locate probability “in the world.” Here, p characterizes the physical propensity or disposition of a given type of physical situation to yield event e or to yield a long-run
relative frequency p of such an outcome-type. To say that e has an objective probability p
of occurring is thus to imply, minimally, that it is in fact possible that e will occur. For a
seminal discussion of the distinction between epistemic and objective probability, see
David Lewis, “The Subjectivist’s Guide to Objective Chance,” in Philosophical Papers, vol.
2 (Oxford: Oxford University Press, 1986).

198

Philosophy & Public Affairs

a gamble was, in a sense, always going to lose it. In terms of objective, if
not of epistemic, probability, there was no chance that the gamble would
turn out in her favor.
From this angle, it may now seem that the problem with imposing
social risks is not just, as the Argument from Certain Loss assumed, that
the agent knows with statistical certainty that someone will end up losing
if she imposes the risk. Rather, the agent knows that whoever ends up
losing was in fact certain to lose, in objective terms. But surely, the
argument goes, a principle allowing us to condemn someone to an objectively certain loss, so that others may enjoy smaller individual benefits, is
one that the loser could reasonably reject. On this view, there is an
important difference between cases of social risk such as Mass Vaccination (Unknown Victims) and single-person gambles like Individual Vaccination: In cases of social risk, we know with statistical certainty that
there will be some losers (by the Law of Large Numbers and the assumption that individual outcomes are probabilistically independent). Moreover, if determinism is true, we know that those who end up losing were,
in fact, objectively certain to lose. In single-person gambles, we also
know that if the person loses, she was objectively certain to lose.
However, unlike situations of social risk, we have no statistical reasons to
believe that we are in a case in which someone will lose. Indeed, in
Individual Vaccination, the available evidence makes it rational for us to
believe, with credence 0.999, that Clara will win her gamble (indeed, that
she is objectively certain to do so). This may make it permissible to give
her Vaccine 2, even if doing the same in Mass Vaccination (Unknown
Victims) would not be justifiable to each person. Call this the Argument
from Determinism.
If sound, the Argument from Determinism would have strongly revisionary implications. Ex ante reasoning could continue to be employed
for social risk-imposition under conditions of causal indeterminacy.
However, in all domains in which the assumption of causal indeterminacy is not plausible (and this, I have said, may include most risky
human activity), contractualist decision makers who cared about
making their actions justifiable to each person would have to resort to an
ex post model of justification. We have already seen that this would have
a profoundly confining effect on the conduct of social policy, which often
involves an ineliminable element of risk. Given these stakes, we should
take a hard look at the Argument from Determinism to examine whether

199

Contractualism and Social Risk

the fact of causal determinism really has the profound normative significance that this argument attributes to it.
In so doing, it is helpful to first go back to a single-person case like
Individual Vaccination, and to ask whether, in that situation, the presence or absence of objective chance makes any difference to Clara’s
reasons for wanting to take Vaccine 2. Let us distinguish two scenarios: in
the indeterministic scenario, the success of Vaccine 2 is objectively
chancy, contrary to what I have been assuming so far. That is, given
Vaccine 2, Clara has a 999/1000 objective chance of surviving the virus
unharmed and a 1/1000 objective chance of dying. In the deterministic
scenario, the outcome of Vaccine 2 is merely epistemically chancy. Given
the available evidence, there is a 999/1000 epistemic chance that Vaccine
2 is objectively certain to allow Clara to survive the virus unharmed, and
a 1/1000 epistemic chance that she is objectively certain to die if we give
her that medicine.
Now, I cannot see any good reason why Clara should hope to be in the
indeterministic rather than the deterministic scenario. In both situations, Clara’s rational degree of credence that she will actually avoid
death is the same. And from the perspective of prospective self-interest,
this is all that appears to matter. (If you are not yet persuaded, the
following analogy might help: All else equal, is there any greater selfinterested reason to buy a ticket for a lottery where the prize will be
allocated by a genuinely indeterministic mechanism [a “quantum randomizer”] than to buy a ticket for a lottery with the same epistemic odds
where a ticket is either sure to win or sure to lose, in the objective sense
[a scratch-card lottery with a preagreed winning number]? Again, it is
hard to think of any such reason. If you had a ticket to the indeterministic
lottery, and someone offered to trade it for an equivalent ticket to the
deterministic lottery plus a small amount of cash, there would be no
good reason to refuse this offer.)
Nor could it be said that if the gamble goes badly, it is worse for Clara
to become disadvantaged as the result of a causal process that was deterministic as opposed to indeterministic. The prudential value of receiving
a chance of some good, whether epistemic or objective, is strictly parasitic upon the value of the good itself. It is no benefit, in and of itself, to
have had an objective rather than a merely epistemic chance of receiving
the good. If, therefore, the gamble does not end up giving Clara the
benefit she sought, it would be in no way better for her to have “at least”

200

Philosophy & Public Affairs

had an objective rather than a merely epistemic chance of obtaining the
benefit. Once we hold fixed the fact of Clara’s death, there is no retrospective reason for her to care whether this happened due to a process
that was objectively or merely epistemically chancy.
I conclude that there is no reason, neither a prospective nor a retrospective one, why a single patient like Clara should care whether she is in
the deterministic or the indeterministic scenario.
These arguments carry over to the case of social risk. Here, too, the
only thing each child has reason to care about is whether he is being
treated in a way that will allow him to actually survive the disease
unharmed. But, for each child, our rational degree of credence that
Vaccine 2 will actually allow that child to survive unharmed is just as high
in the deterministic as in the indeterministic scenario.
Of course, we know that not all children will, in fact, survive if we
choose Vaccine 2. Some will die because the vaccine fails to protect
them from the virus. But this is due to the Law of Large Numbers, paired
with the assumption that individual outcomes are probabilistically
independent, not the issue of determinism vs. indeterminism. Moreover, as I argued in discussing the Argument from Certain Loss, this
fact does not increase any child’s personal reasons for rejecting a principle licensing us to select Vaccine 2. To claim it does would be to
engage in the interpersonal aggregation of claims across different possible worlds.
All told, it is hard to see how the fact that we can predict the overall
pattern of outcomes in Mass Vaccination (Unknown Victims) would give
anyone a greater harm-based complaint against Vaccine 2 than in a
single-person gamble. I conclude that the Argument from Irrelevant
Information fails. By assimilating cases of social risk-imposition to interpersonal trade-offs under certainty, the ex post contractualist conflates
the moral significance of “we know that someone will be harmed” with
that of “there is someone who we know will be harmed.”
Let us take stock: if my arguments for ex ante and against ex post
contractualism have been successful so far, I have shown how the
epistemically chancy nature of many public-policy decisions may allow
a contractualist to justify the imposition of social risks that will
foreseeably result in a pattern of benefits and burdens that his theory
would have forbidden him from bringing about through an interpersonal trade-off under certainty.

201

Contractualism and Social Risk

A catchy way of summarizing the appeal of ex ante contractualism is
that it permits us to “count the numbers without aggregating”: For a
population of a given size, the greater the number of people who would
foreseeably experience some harm H as the result of some socially risky
action, the worse the prospect that this action gives to each individual
(and mutatis mutandis for benefits). In this sense, an ex ante view
“counts the numbers,” because the number of those who will end up
being harmed by the action is an indicator of the prospect that it gives to
each individual. At the same time, ex ante contractualism does not
aggregate, because the structure of justification remains individualistic.
Overall outcomes matter because of what they say about individual prospects, not because of their total goodness or the combination of claims.

VI. EX ANTE RULES AND THE DECOMPOSITION TEST

The intermediate result I have reached stands in sharp contrast with the
position defended by T. M. Scanlon in What We Owe to Each Other.
There are a number of passages in which Scanlon explicitly rejects ex
ante contractualism. He writes:
In considering whether a principle could reasonably be rejected we
should consider the weightiness of the burdens it involves, for those
on whom they fall, and the importance of the benefits it offers, for
those who enjoy them, leaving aside the likelihood of one’s actually
falling in either of these two classes.28
Scanlon concedes that it is “intuitively obvious that the likelihood that a
form of behavior will lead to harm is an important factor in determining
its permissibility.” However, he argues that this does not require that we
“take this probability into account . . . as a factor that, in one way or
another, diminishes the complaint of a person who suffers this harm.”
Rather,
the probability that a form of conduct will cause harm can be relevant
not as a factor diminishing the “complaint” of the affected parties
(discounting the harm by the likelihood of their suffering it) but rather
28. Scanlon, What We Owe to Each Other, p. 208.