This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
It is sometimes permissible to impose a risk of harm on others. Consider,
for instance, driving.1 Intuitively, it is permissible to drive, and permit-
ting this is socially beneficial. The activity does, however, impose a non-
trivial risk of serious harm on both drivers and nondrivers. Requiring
drivers to comply with safety standards significantly lowers that risk, but
it cannot be entirely eliminated. And over time, it is certain to eventuate
in both injuries and deaths.
Work on this article was supported by the Social Sciences and Humanities Research
Council of Canada. Earlier versions were presented at the University of Toronto, the
Arizona Normative Ethics Workshop, McGill University, the University of Oxford, Univer-
sity College London, the University of St. Andrews, Pompeu Fabra University, the Univer-
sity of Graz, Boston University, Queen’s University, the University of British Columbia, and
Claremont McKenna College. I am grateful to the audiences at each of these occasions for
their questions and the discussion. For particularly helpful conversations or comments on
earlier drafts, my thanks to Guy Fletcher, Johann Frick, Barbara Fried, Joe Heath, Louis-
Philippe Hodgson, Michael Gibb, Grant Lamond, Dave Langlois, Andrew Lister, AlistairMacleod, Lukas Meyer, Andrew Ross, Pranay Sanklecha, David Silver, Angie Smith, Nic
Southwood, Daniel Starr, Alan Strudler, Gerard Vong, Andrew Williams, Jo Wolff, and two
anonymous referees for Philosophy & Public Affairs . I am especially indebted to John
Oberdiek for conversations about this topic that both sparked my interest in and helped
shape my thinking about it, and to Kerah Gordon-Solmon, whose extensive and insightful
comments on several earlier drafts vastly improved every page.
. I will treat the risk of some event occurring as the probability of that event occurring
multiplied by the potential harm. For present purposes, nothing hangs on the choice
between characterizing risk this way rather than simply as the probability of an event
occurring. I will also assume an epistemic understanding of probability as what is relevant
in the context of determining what to do or expect. A useful general discussion of different ways in which “risk” is characterized in the literature is Madeleine Hayenhjelm and Jona-
than Wolff, “The Moral Problem of Risk Impositions: A Survey of the Literature,” European
Justifying the permissibility of this kind of socially productiveactivity is straightforward in broadly consequentialist cost-benefit
terms: the expected aggregate benefits over time of permitting driving,
subject to certain constraints, outweigh the expected aggregate
burdens. The question I will pursue here is whether, and if so, how, the
permissibility of this kind of risk-imposing activity can be justified
in nonconsequentialist terms.
Nonconsequentialists, of a broadly Kantian stripe, take consider-
ations of aggregate benefit and burden to have no bearing on the justi-
fication of the permissibility of conduct. But the permissibility of most
intuitively acceptable risk-imposing activities cannot, it appears, be
made sense of without recourse to such considerations. This claim is
persuasively argued for by Elizabeth Ashford with respect to Scanlon’s
nonconsequentialist moral theory, contractualism.2
Assessing the permissibility of driving in contractualist terms is,
roughly, a matter of comparing the force of the reasons someone might
have for wanting driving to be permitted (such as the convenience and
freedom it affords) with the reasons a person could offer, on her ownbehalf, for wanting it to be prohibited (such as not wanting to be killed as
a result of a car accident). Looking at it in this way, it is hard to see how
the permissibility of driving can be defended. Though there are many
respects in which a person stands to benefit from driving, in some form,
being permitted, they all inevitably pale in comparison to the burden
of being killed.
In this article, I will argue that contractualist reasoning does not
support radical, and implausible, conclusions concerning permissiblerisk imposition. On the contrary, properly understood, the approach
offers an attractive nonconsequentialist framework for assessing the per-
missibility of a risk-imposing activity.
. Elizabeth Ashford, “The Demandingness of Scanlon’s Contractualism,” Ethics
(): –. A complementary critique of contractualism that reaches a similar con-
clusion is advanced by Barbara H. Fried, “Can Contractualism Save Us from Aggregation?” Journal of Ethics (): –. See also Barbara H. Fried, “What Does Matter? The Case
for Killing the Trolley Problem (Or Letting It Die),” Philosophical Quarterly (): –
. In this discussion, I shall focus solely on Ashford’s articulation of the case for why
intuitively plausible conclusions concerning permissible risk imposition cannot be made
In this section, I will briefly lay out the contractualist account of what
makes conduct impermissible. This will provide the background needed
to motivate the claim that permissible risk imposition cannot plausibly
be made sense of in contractualist terms.3 I will say more about the
account and, in particular, about how the permissibility of a risk-
imposing activity is to be assessed in its terms in the next sections of the
article responding to Ashford’s objections.
Contractualism says that an act is impermissible if any principle per-mitting it could be reasonably rejected by an appropriately motivated
individual; an individual is appropriately motivated if she is moved to
find principles for the general regulation of conduct that no one, simi-
larly motivated, could reasonably reject. Regulating one’s conduct in a
way that conforms with such principles is what respect for the value of
persons, as beings capable of assessing reasons and governing their lives
accordingly, requires.4 A person is wronged when another’s conduct
toward her either intentionally or negligently flouts these requirements, which flow from her standing as a person to whom justification is owed.5
A principle for the regulation of a certain type of conduct that no one
(appropriately motivated) can reasonably reject is one that all individu-
als have reason, as assessed from each person’s own point of view, to
license one another to be regulated by. Whether a proposed principle is
one no one can reasonably reject turns on comparing its implications to
the implications of candidate alternative principles, as assessed from
. The relevant sense of “morally wrong” or “morally impermissible” presupposed
throughout this discussion is what Parfit calls the evidence-sensitive sense of wrong.
It ties what it is morally permissible to do to the available evidence at the time of acting,
or what one ought to believe it is permissible to do given the available evidence, and our
beliefs being true. See Derek Parfit, On What Matters , vol. (Oxford: Oxford University
Press, ), sec. . It is the sense of wrong presupposed by the contractualist account,
and in a deterministic world, the relevant sense of wrong for understanding the morality
of risk imposition.
. T. M. Scanlon, What We Owe to Each Other (Cambridge, Mass.: Harvard University
Press, ), p. . The distinction between acting for the reasons identified by a relevantprinciple as conclusive reasons for so acting (complying with reasons) and conducting
oneself in the way owed to others though not for those reasons (conforming with reasons)
is relevant here. Wronging another only requires a failure to conform. A failure to comply
with the relevant principle bears on the blameworthiness of one’s conduct.
If a principle is one no onecan reasonably reject, the strongest objection to it will not be as strong as
those that can be pressed from other points of view against every plau-
sible alternative. It is the principle whose implications are most accept-
able to the person to whom it is least acceptable.
Importantly, only the implications of a principle that bear on an indi-
vidual being able to lead her life are relevant to the assessment of its
reasonable rejectability.7 Impersonal or nonpersonal considerations,
such as those concerning aggregate benefit and burden, have no role to
play. Set aside, then, are the considerations that a cost-benefit approach
takes into account in assessing a risk-imposing activity’s permissibility.8
Whether, and on what terms, it is permissible to do what imposes a
risk of harm on others depends on whether a principle that permits
. This involves understanding “alternative principle” as some other principle for the
general regulation of the type of situation in question that performs roughly the same
function as the original principle.
. This is what has, following Parfit, come to be known as the “individualist restriction.”See Scanlon, What We Owe to Each Other , pp. –; and Parfit, On What Matters , vol. ,
sec. ; Rahul Kumar, “Contractualism on the Shoal of Aggregation,” in Reasons and Rec-
ognition: Essays on the Philosophy of T. M. Scanlon, ed. R. Jay Wallace, Rahul Kumar, and
Samuel Freeman (Oxford: Oxford University Press, ), pp. –; and Rahul Kumar,
“Reasonable Reasons in Contractualist Moral Argument,” Ethics (): –.
. Setting aside aggregate considerations as relevant to the justification of what
it is permissible to do commits contractualism to a “probability” (or “individual”)
rather than a “frequency” (or “population”) perspective on risk. Barbara Fried nicely
illustrates the distinction:
Suppose we estimate that if we distribute a new Flu vaccine, roughly one out of every one million people inoculated will have an adverse reaction, resulting in death. We plan
to inoculate million people. There are two ways to describe the likelihood that death
will result from the inoculations, sometimes differentiated as “probability” versus “fre-
quency.” The first (“probability”) is the odds that any given person who receives the
vaccine will die (one in a million). The second (“frequency”) is the total number of
expected deaths if we inoculate million people (ten). Both are describing the same set
of acts, predicted to cause the same consequences with the same likelihood; they are
simply describing the likelihood from different perspectives. (Fried, “Can
Contractualism Save Us from Aggregation?” p. )
Contractualism takes what is relevant to the assessment of the permissibility of a risk-
imposing activity to be whether the permission is justifiable to the person who stands to be
most at risk if it is permitted (invoking a probability perspective on risk). On a cost-benefit
approach, what matters is the likely number of individuals who will end up harmed if the
activity is permitted on the proposed terms (invoking a frequency perspective on risk).
doing so is justifiable to each person as one no one can reasonably reject.Focusing on what is justifiable to each turns out, however, to make
justifying the permissibility of risk-imposing activities more difficult
than it is on a cost-benefit approach. Consider, for example, the permis-
sibility of introducing a product similar to peanut butter to the market-
place. Doing so, assessed in cost-benefit terms, looks to be permissible.
Though the benefit to any given individual of it being available on super-
market shelves is small, the aggregate benefit is enormous. There is a risk
that some will either become seriously ill or die as a result of an allergic
reaction to it, but that risk (understood as the expected number who will
end up ill or dead over time as a consequence of the permission) can be
kept low through the use of warning labels and other precautions.
A contractualist assessment of the permissibility of introducing it, on
the other hand, asks: what is the risk of ending up harmed that will be
imposed on any individual (especially the most at risk) by the product
being available, and why might it be important to an individual that it be
permissible to make it available? It supports the opposite conclusion to
that supported by cost-benefit considerations: it is hard to see how anindividual’s loss of the benefit from access to the peanut-butter-like
product could be of sufficient importance to justify to one allergic to it
the imposition of even a low risk of death.9
Contractualist reasoning may support some surprising conclusions
concerning the permissibility of risk-imposing conduct. The worry is
that it supports the prohibition of all risk-imposing activities. Ashford’s
case for concluding that it does is as follows: consider the justifiability of
. In some cases, taking what is relevant to the assessment of a risk-imposing activity to
be the risk imposed on an individual (a probability perspective) might be thought to lead to
implausible conclusions. Say you are faced with a choice between imposing a large risk of
injury on those characterized by a certain standpoint that in fact happens to characterize
very few existing individuals and imposing a smaller risk on another standpoint that
happens to characterize a very large number of existing individuals. Many will be inclined
to think that the correct choice is to impose a greater risk on the smaller number. That way,
over time, you are just going to end up with fewer injured people.
Here, I will just note that the issue raises no special questions that arise in the context
of thinking about the grounds for permissible risk imposition. Rather, it concerns thegeneral question of how the contractualist approach makes sense of the seeming relevance
of aggregative considerations for intuitive convictions concerning what, in certain cases, it
is permissible to do. For discussion of the issues, see Aaron James, “Contractualism’s (Not
So) Slippery Slope,” Legal Theory (): –; and Kumar, “Contractualism on the
to one who knows she is to be experimented on against her will. Thebenefits secured for individuals by permitting it would have to be espe-
cially important to even start to make it plausible that it might be.
Requiring that the permissibility of conduct be justifiable to the one
who stands to have to bear the greatest burden if it is permitted, and
knows that, appears to yield a plausible rationale for the prohibition of
the experimentation. But now consider a second case, that of an Amish
farmer, Jebidiah, who lives under a heavily used flight path (fixed by a
process that did not involve consulting those who live in its vicinity).12
While out tilling his fields, he is hit by a bit of falling airplane fuselage and
is fatally wounded.13 Stipulating the risk of his being injured to be small
and the benefit secured for others by routing flight paths over Amish
country as quite significant (I will assume it would be financially unfea-
sible otherwise), it is plausible to hold Jeb’s injury to be the result of an
activity’s permissible pursuit. But it is hard to see how the permissibility
of the aviation practice in question could be defended as justifiable to a
fatally wounded Jeb as something no one, including him, can reasonably
reject. What could one say to him? “You understand: though it has cost you your life, the alternative to not permitting commercial flights is to
ask others each to individually bear a much greater burden”? It is
implausible to suggest that, in a comparison of the burden of not being
. This case is first introduced in Ashford, “The Demandingness of Scanlon’s
Contractualism.” I introduce it here because it is one that has gained traction in the
discussion of this topic. See, for instance, V. Munoz-Dardé, “Global Justice: Imposed and
Shared Risks,” in Spheres of Global Justice , Vol. , Fair Distribution—Global Economic,
Social and Intergenerational Justice , ed. Jean-Christophe Merle (Dordrecht: Springer,),
pp. –, a discussion of contractualist and risk imposition complementary to this one. It
is worth noting, however, an important respect in which the case is misleading. The Amish
are thought to eschew the use of modern technology, so it looks like the Amish farmer does
not stand to benefit in any way from the permitting of commercial aviation on the pro-
posed terms. The intuition the case is meant to elicit is that this makes its permissibility
particularly difficult to justify to him. But permitting an activity need not be mutually
beneficial for it to be true that its permissibility is justifiable to each person. What mattersis that it be one that individuals have good reason to want to be able to permissibly pursue,
and that the burdens some individuals might each have to bear if it is permitted be not so
great as to make accommodating its pursuit unreasonably burdensome.
. These things do happen: http://abcnews.go.com/blogs/headlines///plane
Inthissection,Iwillarguethatthetwocasesarenot,infact,structurally identical. Contractualist reasoning does yield a plausible rationale for
the medical experimentation’s impermissibility. But it has no implica-
tions for the commercial aviation case. (I will return to the question of
what contractualism has to say about that case in the next section.)
The seemingly compelling case for prohibiting the medical experi-
mentation identifies the decisive objection to it in the burden imposed
on one actually experimented on. Comparing it to the force of any indi-
vidual objection to it not being permitted, it is clear that no one stands to
be as burdened by its prohibition as by it being permitted. A slight modi-
fication to the case, however, casts doubt on whether this really is a
compelling rationale for its impermissibility. Say the experimentation is
still involuntary, but it involves painlessly enhancing individuals in ways
that significantly benefit each of them. Further, it takes place without the
experimental subjects’ knowledge, by slipping drugs into the water
supply of their homes. Without the element of pain or distress at the
knowledge that one is being involuntarily experimented on, and no
prospect of ending up burdened as a result, the proposed rationale for itsimpermissibility ceases to look so compelling.15
This rationale for not permitting the experimentation locates the deci-
sive objection to it in what I will call an instrumental consideration, one
that concerns a respect in which an individual stands to be benefited or
burdened as a result of an activity being permitted.16 But, intuitively,
whether the involuntary experimentation is painful and harms those
subjected to it, or is painless and results in their having better lives, has
little bearing on whether or not it is permissible. An alternative rationale,one that does justice to this thought, stresses the relevance to the case of
. I assume here that the enhancement is not burdensome because it is involuntary.
. Ashford’s argument assumes that, in assessing a principle’s reasonable rejectability,
what is relevant is the undiscounted burden an individual might have to bear as a result of
the envisioned permission. As she sees it, allowing a burden to be discounted by the
probability of a person having to bear it would undermine the compelling contractualist
rationale for the medical experimentation’s impermissibility. In Section IV, I will argue that
Scanlonian contractualism is best understood as taking the discounted, not the full, poten-tial burden that an individual might have to bear to be what is relevant in assessing the
strength of the person’s objection to a proposed principle. In this section, I will simply
grant Ashford’s assumption. I will argue, instead, that the issue of whether or not potential
burdens are to be discounted has no bearing on what contractualist reasoning leads us to
conclude about the permissibility of the medical experimentation.
sucked from the ground into an airplane by a giant vacuum and thenejected out of the cargo bay, from which she falls to the ground and is
either badly injured or killed. In this version of the case, to permit com-
mercial flights is to permit an activity that grants discretion to make
decisions concerning how an individual’s body is used to people other
than the individual herself. The activity is one, after all, whose success
requires that it be permissible to suck an individual off the ground and
into the plane, without her consent, as if she were a form of fuel for the
flight. This contrast brings into relief that the permission in question in
the original version of the case is not one that involves any such usurpa-
tion of an individual’s authority to make decisions concerning how her
body is used.
IV
The medical experimentation and commercial aviation cases are not, as
Ashford contends, symmetrical in structure. But that does not show her
main claim, that contractualist thinking supports implausible conclu-sions concerning the permissibility of risk-imposing activities, to be
mistaken. The most plausible grounds for taking commercial aviation to
be impermissible are instrumental in character.19 Any instance of a
flight over inhabited territory imposes a risk of harm on those inhabit-
ants. Over time, some of them will be unlucky enough to be harmed as
a result of that risk eventuating. The question to be asked in assessing
the activity’s permissibility is whether the risk of harm imposed by
commercial flights on those who just happen to live under a flight
path is justifiable to any such person on grounds that person cannot
reasonably reject.20
. I am assuming that the Amish would not find the very presence of an aircraft flying
over their territory an affront to their religion or way of life, so I am setting aside that
consideration as grounds of a possible intrinsic objection.
. It could be argued that the familiar distinction between intended and foreseen
consequences accounts for the asymmetry between the cases. Though individuals ending
up being harmed is a foreseeable consequence of commercial aviation being permitted, it
is not intended. In the medical experimentation case, on the other hand, there is anintention that people (though no one in particular) be harmed. This difference makes the
permissibility of aviation prima facie more defensible than the medical experimentation.
For present purposes, whether or not this is a morally relevant difference between the
cases is not an issue on which I need take a stand. What I am arguing is that it is the
distinction between intrinsic and instrumental considerations that is load-bearing in how
Because the permission is a generalpermission to act for anyone thusly situated, assessing whether it is
permissible to act, here and now, in a way that has implications for
a particular other (or others), requires considering the general question
of what the reasons are that favor and oppose this type of conduct
being generally permitted.
The reasons that bear on this question concern the relevant intrinsic
and instrumental considerations that could be appealed to from the
different points of view of those who stand to be affected. Because what
is licensed by a principle is the general permissibility, in a certain type of
situation, of relating to another in a certain way, and there is no way of
knowing how often it will obtain or who will find themselves in it, the
points of view in question are not those of actual, particular individuals,
but rather representative individual standpoints.
A “standpoint,” as the term is being used here, is an abstraction, a way
of referring to the reasons that persons in certain circumstances
(assumed to have certain “normal” capacities for discernment, self-
control, planning, and so on) typically have for caring about or wanting certain things (“generic reasons”), such as a degree of personal privacy,
security against bodily injury, the freedom to nurture personal relation-
ships with friends and intimates, and sole discretion concerning how
one’s body is used.22 The relevant standpoints for assessing principles
governing promissory obligations, for example, are those of the promisor
and the promisee, each of which is associated with certain characteristic
interests—such as the promisee’s interest in being assured that certain
things will or will not be done (unless she says otherwise), and the promi-sor’s interest in being able to assure the promisee, if she seeks assurance.23
. Scanlon, What We Owe to Each Other , pp. –.
. A generic reason is “one that we can see people have in virtue of certain general
characteristics; it is not attributed to specific individuals.” Scanlon, What We Owe to Each
Other , pp. –; see also T. M. Scanlon, “The Significance of Choice,” Tanner Lectures on
Human Values , pp. –. It is worth noting that the notion of a “generic reason” is a
wholly normative claim, one having to do with the reasons that individuals in certain
circumstances normally have for wanting or caring about certain things. That some peoplein the actual world may not in fact have reason to care about these things does not call
reliance on generic reasons into question, as they are not statistical generalizations based
on a particular population.
. In assessing the implications of a proposed principle for a particular standpoint,
whose particular interests are aptly characterized by that standpoint is of no importance.
Relying on standpoints is motivated by the thought that conclusionsabout the permissibility of conduct must be reasonably epistemically
accessible to an individual thinking about what courses of action are
open to her. Requiring that an individual take into account specific facts
about the rational preferences, capacities, psychological dispositions,
and so on of others in determining what it is permissible to do would
place serious epistemic and cognitive demands on a person. This would
constrain an individual’s ability to pursue her rational aims, making it
unduly burdensome for her to live her life within the bounds of the
permissible, and make it unreasonably difficult for both her and others
assessing it to be confident of her conduct’s permissibility. These prob-
lems are avoided by relying, in thinking about what it is permissible to
do, on relevant individual standpoints, characterized in terms of generic
information about what individuals in the type of situation in question
typically have reason to want.
With these last points concerning the role of principles and stand-
points in contractualist thinking in mind, let me restate how the question
of the permissibility of a risk-imposing activity is framed in its terms: thereasonable rejectability of a principle for the regulation of a risk-
imposing activity turns on a comparison of the implications of conduct
being regulated by that principle with those of candidate alternative
principles, as assessed from the relevant standpoints. The implications
in question are either intrinsic or instrumental considerations. For any
proposed principle, there will be numerous such considerations that
favor permitting the activity on its terms. The main objection to doing so
will most often have to do with the generic reason a person has for not wanting to end up burdened as the result of the imposed risk eventuating
in her life.
How forceful an objection this is will depend on how likely it is that
permitting the activity will result in a person ending up burdened.24 The
Any standpoint could (especially over time) be predicated of an indefinite number of
particular persons. What matters for purposes of fixing what the relevant standpoints are
for assessing a proposed principle are, first, that a standpoint aptly characterize the rel-evant interests of an individual or individuals who, in the world as we know it, exist, did
exist, or we have reason to believe will exist, and second, that it be one for which the
proposed principle has implications.
. I am in agreement with Lenman in seeing principles as introducing an ex ante point
of view from which the reasonable rejectability of a principle is to be assessed. Unlike
magnitude of the risk imposed on different standpoints, and, especially,the standpoint of those most at risk, will therefore be an important deter-
minant of its permissibility. But, for reasons I will briefly discuss below,
any legitimate answer to this question is best understood as correct
relative to certain assumed parameters that can be specified in numer-
ous different ways.25
Consider a common understanding of probability, the relative fre-
quency view. It holds that the probability of a certain event-token occur-
ring is fixed by first defining some reference class that contains the object
of risk assessment, and then determining what proportion of the people
or events in that class have the relevant property. For example, to deter-
mine what the risk is of my being injured playing squash over the next
decade, a reference class needs to be defined, and the proportion of
individuals in that reference class who have been injured playing squash
needs to be determined. Casting the proportion of individuals in the
reference class who exhibit the property of interest (injured while
playing squash over a certain time span) in terms of frequency tells me
what the probability is of my being injured in that time period. What exactly that turns out to be depends on how the reference
class is characterized, for which there are a myriad of possibilities. A
candidate reference class need only pick out some property that can be
correctly attributed to me and is also instantiated by others. It could,
for instance, be the class of men, men of a certain age, or squash
players with a certain number of years of experience playing squash.
There are, in principle, as many legitimate answers to the question of
Lenman, I do not believe the magnitude of a burden that may befall a person if conduct of
a certain type is permitted, as opposed to the magnitude of that burden discounted for the
probability of it befalling a person, to be relevant to the assessment of whether it is in fact
permissible. See James Lenman, “Contractualism and Risk Imposition,” Politics, Philoso-
phy, and Economics (): –. Fried, “Can Contractualism Save Us from Aggrega-
tion?” draws attention to certain passages in Scanlon, What We Owe to Each Other , that
suggest that undiscounted burdens are relevant to assessing a principle’s reasonable
rejectability. My view is that these passages are misleading. Contractualism is best under-
stood as holding that only instrumental and intrinsic considerations are relevant to assess-
ing a principle’s reasonable rejectability and that undiscounted burdens are not relevant.. The next few paragraphs largely follow points elegantly made in Alan Hàjek, “Con-
ditional Probability Is the Very Guide of Life,” in Probability Is the Very Guide of Life: The
Philosophical Uses of Chance , ed. E. Henry Kyburg Jr. and Mariam Thalos (LaSalle, Ill.:
Open Court, ), pp. –; and Alan Hàjek, “The Reference Class Problem Is Your
what the probability is of my being injured while playing squash in thenext ten years as there are reference classes.26 Each legitimate answer
to the question is going to be fixed relative to the parameters of the
assumed reference class.27
. I have relied, in making this point, on a relative frequency interpretation of prob-
ability. But the point can also be made on the terms of the other major interpretation of
probability, the Bayesian view. The Bayesian view holds that the risk of a certain event
occurring is to be fixed relative to a suitable agent’s degrees of belief, or confidence, in aproposition stating the probability of an event occurring. On the simplest Bayesian view,
any proposition stating the probability of an event occurring is always relative to a particu-
lar agent’s degrees of belief. There are potentially as many legitimate answers to the ques-
tion of how great the risk imposed by a certain activity is as there are agents.
It is more plausible to hold that the probability of an event occurring ought to be fixed,
not with respect to just any agent’s degrees of belief, but with respect to the degrees of
belief of those with suitable expertise and knowledge. Doing so, however, in no way under-
mines the conclusion that there are many valid answers to the question “how risky is it?” As
Alan Hàjek points out, an expert meteorologist may say that there is a percent chance of
tomorrow being a rainy day because days like tomorrow have been rainy days. But there are
innumerable ways that tomorrow might be characterized as being “like” other days. Whatthe probability of tomorrow being a rainy day turns out to be will be conditional on how the
“like” relation is filled in; there are lots of reasonable possibilities here. Any expert will have
to choose among them, and there is no reason to think that experts will not differ in the
choices they make.
. There is no fixed rule for picking out one reference class as the appropriate one.
Something like “choose the narrowest possible relevant reference class for which reliable
evidence is available” leaves open the choice between roughly equally narrow reference
classes, and requires that some kind of standard of what counts as “reliable” be employed.
Similarly, a rule that directs us to fix what the risk is by appeal to the degrees of belief of an
expert leaves open to judgment what is going to count as expertise for the purposes of fixing
the kind of risk in question, and how to decide which expert’s view to go with in the familiar
circumstance that experts on the matter in question disagree among themselves.
It could be argued that this reference class problem does not show that, as a metaphysi-
cal matter, there is no such thing as the unique probability of a proposition in virtue of the
complete set of conditions that hold in the universe at a particular time. For purposes of
this discussion, it is enough to take it to show that there is a problem with fixing the
probability of a proposition (or event-token) in a form that is informative for purposes of
deciding what to do or expect, while retaining a claim to it being the unique, correct
answer. As Hàjek puts it: “Consider again poor old John Smith’s predicament at this
moment [he is a fifty-year-old consumptive Englishman, wondering what his chances are
of living to sixty-one]. Imagine him having knowledge of the complete situation of theuniverse at this moment; or knowledge of a complete set of relevant conditions. We tell him
that propensities are dependent on these things, but we do not tell him how. Granting him
all the computational power that he might need, does he have any idea what is his pro-
pensity for living to , or even what this means?” Hàjek, “The Reference Class Problem Is
The information required for characterizing as distinct, and relevantto the assessment of the vaccine program’s permissibility, the stand-
points of those who stand to benefit from it, but who differ in the degree
to which they are put at risk by being vaccinated, is certainly available.
But the burdens associated with obtaining the relevant information,
together with the fact that doing so is in tension with mounting an
effective vaccination program in a timely manner, provide grounds for
resisting assessing its permissibility from multiple standpoints that differ
with respect to a susceptibility to an allergic reaction. Arguably, were
they to be taken into account, there would be a good case to be made that
testing for the genetic mutation is one of the conditions of permissibly
carrying out the program; as it is, it is permissible to proceed with the
vaccination program without testing for the mutation. If an individual
does suffer an allergic reaction, she will be owed assistance, but cannot
claim to have been wronged by having been exposed to the risk of this
happening to her.28
B
The argument for why contractualism supports implausible conclusions
about permissible risk imposition relies on the thought that most intui-
tively acceptable risk-imposing activities will not be justifiable as per-
missible to someone like Jeb, who knows that the risk in question has
actually eventuated in his life as a serious harm. But whether Jeb has
been wronged by having that risk imposed on him turns on whether
doing so is permitted by a principle that is justifiable to anyone (suitably motivated) on grounds that no one, including Jeb, can reasonably reject.
. In this example, the epistemic demands associated with different ways of charac-
terizing relevant standpoints count in favor of some ways of characterizing a standpoint
over others, and thus function as grounds for privileging one way of characterizing the
riskiness of an activity for purposes of assessing its permissibility. But it is not the only type
of consideration that might play this role. In the vaccination program case, for instance,
there is a distinct kind of objection to taking into account, in assessing its permissibility, the
risk to those with a specific gene mutation, namely, that gathering such genetic informa-
tion is likely to involve nontrivial intrusions into individuals’ lives. Arguably, individualshave an interest in having their private information—especially their medical
information—protected from official authorities (like government agencies). Gathering the
information required to come up with a reasonably reliable estimate of the risk to which
individuals with the mutation stand be exposed by the vaccination program will almost
Answering this question requires taking up a kind of ex ante point of view. It is not ex ante in the sense that the question to be asked is what
principle would be justifiable to Jeb if he did not know his particular
identity and that he has been fatally wounded. Rather, it is an ex ante
point of view in the sense that it only takes into account a principle’s
instrumental and intrinsic implications, as assessed from the relevant
standpoints. Jeb knows that he has ended up seriously harmed as a result
of another’s conduct. But whether or not he has been wronged turns on
whether the risk that the kind of thing that has happened to him could
happen to someone in his circumstances is sufficient grounds to take
any principle permitting such conduct to be reasonably rejectable. That
Jeb has in fact been harmed is not in itself grounds for concluding
that any principle that permits the aviation practice in question is one he
can reasonably reject.
It seems plausible to take the risk of ending up harmed as a conse-
quence of an aviation accident imposed on someone in Jeb’s circum-
stances to be quite low (aviation accidents happen, but not that often).
But as previously discussed, taking the imposed risk to be low assumes acertain characterization of the standpoint of “someone in Jeb’s circum-
stances.” The assumption, roughly, is that the standpoint is one of
someone who lives in a sparsely populated territory that lies under
a flight path.29
Say, however, that Jeb is the sort of person who is temperamentally
drawn to planes. When he hears one passing, he finds himself moved to
try to situate himself so as to be able to get a good look at it. Were this
fact about his temperament incorporated into the characterization of the standpoint of “someone in Jeb’s circumstances,” the risk to which
such a person stands to be exposed would, let us say, be greater than if
it is not taken into account. What justifies not including it are the
epistemic demands associated with discovering whether or not anyone
located in the vicinity of a particular flight path does in fact have Jeb’s
. Say the Amish experience a population explosion, such that their number grows
exponentially, making the territory in question very densely populated. The risk of an Amish person being harmed would then no longer be quite low, as the probability of the
area in the territory in which, for example, a bit of falling aircraft debris lands being
occupied will have gone up substantially. This can be taken to be a respect in which the
numbers are relevant to determining what is permissible, but not in the way ruled out by
always conducting oneself in a way whose permissibility is justifiableto each person.31
VI
In this discussion, I have argued that understanding permissible risk
imposition in contractualist terms allows us to make good sense of the
thought that one person can wrong another by doing what imposes an
inordinate risk of harm on her. This interpersonal aspect of impermis-
sible risk imposition is not one readily captured by a consequentialistcost-benefit approach. I have further argued that it is a strength of the
approach that the important question of how the risk associated with an
activity is to be characterized for purposes of moral deliberation and
assessment explicitly arises in a way that is internal to the process of
working out whether a principle permitting a risky activity can be rea-
sonably rejected. Finally, I have claimed that contractualist reasoning,
though more restrictive than a cost-benefit approach, does not support
an implausibly restrictive understanding of what risks may permissibly be imposed. The case for thinking it does rests on taking the fact that a
particular person has ended up burdened, rather than the risk that a
person might end up so burdened as a result of an activity being permit-
ted, as the grounds for wanting to reasonably reject any principle that
allows it. That, as I have argued, is a mistake.
. The contractualist claim, to be fair, is that realizing the value of mutual recognition
in the moral relationship, one that holds between oneself and any other, requires that one
be guided in one’s conduct by the reasons made salient by principles no one can reason-
ably reject. It is by responding to these reasons, and not others, in one’s practical thinking that one displays a recognition of the standing of others as rational beings to whom
justification is owed in virtue of their value as rational beings. This aspect of the
contractualist ideal of how persons ought to relate to one another has nothing to do,
however, with the issue at hand, that of whether Jeb is wronged by the risk imposed on him