week04_Norcross1997

Comparing Harms: Headaches and Human Lives
Author(s): Alastair Norcross

Source: Philosophy & Public Affairs , Spring, 1997, Vol. 26, No. 2 (Spring, 1997), pp.
135-167
Published by: Wiley
Stable URL: http://www.jstor.com/stable/2961948
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.com/stable/2961948?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Wiley is collaborating with JSTOR to digitize, preserve and extend access to Philosophy &
Public Affairs
This content downloaded from

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
All use subject to https://about.jstor.org/terms
ALASTAIR NORCROSS Comparing Harms:
Headaches and Human Lives
Consequentialists are sometimes unsettled by the following kind of ex-

ample: a vast number of people are experiencing fairly minor head-
aches, which will continue unabated for another hour, unless an inno-
cent person is killed, in which case they will cease immediately. There
is no other way to avoid the headaches. Can we permissibly kill that
innocent person in order to avoid the vast number of headaches? For a
consequentialist, the answer to that question depends on the relative
values of the world with the headaches but without the premature
death, and the world without the headaches but with the premature
death. If the latter world is at least as good as the former, it is permissible
to kill the innocent. Furthermore, if the all-things-considered values of
the worlds are comparable, and if a world with more headaches is,
ceteris paribus, worse than a world with fewer, it is reasonable to sup-
pose that a world with a vast (but finite) number of headaches could be
worse than a world that differs from it only in lacking those headaches
and containing one more premature death. In short, there is some finite
number of headaches, such that it is permissible to kill an innocent per-
son to avoid them. Call this claim life for headaches. Many people balk
at life for headaches. In fact, many people think that there is no number
of people such that it is permissible to kill one person to save that num-
ber the pain of a fairly minor headache. Deontologists might think this,
because they endorse what Scheffler calls "agent-centered restrictions."
Such restrictions forbid certain kinds of action, even when their results
are at least as good as all alternatives. Thus, a deontologist might agree
that the world without the headaches but with the premature death is
Earlier versions of parts of this article were presented at Texas A & M University, the
University of Mississippi, the American Philosophical Association Pacific Division meet-
ing, and the North Texas Philosophical Association. I thank audiences at all these events
for helpful comments. The article has also benefited from the comments of Greg Bayer,
Doug Ehring, David Hausman, Mark Heller, Derek Parfit, and Steve Sverdlik.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
136 Philosophy & Public Affairs
better all things considered than the world with the headaches but with-
out the premature death, and yet maintain that it is impermissible to kill
the person in order to avoid the headaches.' A consequentialist, how-
ever, who agrees with this ranking of the two worlds must also claim that
it is permissible to kill the innocent person.2
In order to avoid life for headaches, therefore, a consequentialist must

deny that a world with a vast (but finite) number of headaches could be
worse than a world that differs from it only in lacking those headaches
and containing one more premature death.3 One reason for such a de-
nial can be stated as follows:
Incomparable: The state of affairs constituted by any number of fairly

minor headaches is incomparable with the premature death of an in-
nocent person.
To say that a state of affairs (or option) A is incomparable with a state

of affairs (or option) B is to say that (i) it is false that A is better than B,
and (ii) it is false that B is better than A, and (iii) it is false that A and B
are of equal (or roughly equal) value.4 There are at least two reasons why
a consequentialist who wishes to avoid life for headaches should not be
satisfied with Incomparable. First, a consequentialist who denies that it
is permissible to kill an innocent person in order to avoid a vast number
1. Of course, it is one thing to claim that such a deontological restriction exists, it is

quite another to argue for its existence. For an attempt to do the latter, see Judith Thom-
son, The Realm of Rights (Cambridge, Mass.: Harvard University Press, 1ggo), Chap. 6. For
an argument that Thomson's approach does not succeed, see Alastair Norcross, "Rights
Violations and Distributive Constraints: Three Scenarios," Pacific Philosophical Quarterly
(June 1995): 159-67.
2. This assumes that agent-neutrality is an essential feature of consequentialism. This
is widely accepted. However, I do not wish to go into the question of whether there could
be a theory that endorsed agent-relativity and is still distinctively consequentialist in na-
ture. This article is addressed to all those consequentialists, which may be all the conse-
quentialists there are, who endorse agent-neutrality.
3. It is not clear what kind of possibility is required to satisfy those who find life for
headaches so counterintuitive. Some may be content with nomological possibility, others
may demand metaphysical possibility. For the purposes of this article, it does not matter.
4. Some philosophers equate incomparability with rough equality. This may involve a
conflation of the notions of incomparability and of vagueness. I believe it avoids confu-
sion to use a notion of incomparability that excludes rough equality as well as equality.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
137 Comparing Harms:
of headaches is most likely to do so because she thinks it worse to kill

than to allow the headaches, not because she thinks the two options are
incomparable. Thus, she will most likely also claim that it is permissible,
in fact obligatory, to allow the headaches to continue, if the only alterna-
tive involves killing an innocent person. But it is hard to see how Incom-
parable would yield the impermissibility of one option and not the
other. Either both options would be impermissible, or neither would be.
This leads to the second problem with using Incomparable to avoid life
for headaches. Whether Incomparable leads to the denial of life for head-
aches depends on which formulation of the consequentialist criterion
of permissibility we employ. Consider the following two formulations:
Con 1: An act is permissible iff it is at least as good as all the available

alternatives.
Con 2: An act is permissible iff there are no better alternatives.
If we accept Incomparable, and consider the situation in which we ei-

ther kill one innocent person or allow a vast number of headaches to
continue, we must still decide which criterion of permissibility to em-
ploy. According to Con 1, the killing is impermissible, since it is false that
it is as good as or better than the only available alternative. Of course,
the alternative, allowing the headaches to continue, is also impermissi-
ble, for the same reasons. In which case, we have a moral dilemma. On
the other hand, according to Con 2, the killing is permissible, since there
are no better alternatives.
II
A more promising tactic for a consequentialist who wishes to avoid life

for headaches is to maintain that the loss of an innocent life is worse
than a vast number of headaches, than any number of headaches, in
fact. Consider the claim in this form:
Less: The state of affairs constituted by any number of fairly minor

headaches is less bad than even one premature death of an innocent
person.
A consequentialist who accepts Less denies that it is ever permissible to

kill an innocent person in order to avoid a number of headaches, no

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
matter how vast. According to Less, if two worlds differ only in that the
first contains a vast number of fairly minor headaches not contained in
the second, and the second contains one more premature death of an
innocent than the first, the second world is worse than the first.
Although Less is initially plausible, it is problematic. Roughly, Less
says that a death is worse than any number of headaches. Consider now
a much stronger principle:
Less': For any misfortune, x, to befall a person, that is worse than that
person suffering the pain of a fairly minor headache, the state of af-
fairs constituted by any number of fairly minor headaches is less bad
than even one person suffering x.
Less' is wildly implausible. Roughly, it says that anything worse than a

headache is worse than any number of headaches. Suppose that a mild
ankle sprain is worse than a mild headache. According to Less', it is
worse that one person suffer a mild ankle sprain than that billions suffer
mild headaches. Clearly, no consequentialist will (or should) accept
this. The rejection of Less' entails that there are misfortunes that are
worse than mild headaches, that nonetheless can be individually out-
weighed by a sufficient number of mild headaches. This is relatively
uncontroversial. A mild ankle sprain is a good candidate for such a mis-
fortune. Likewise, it is pretty clear that there are misfortunes which are
worse than mild ankle sprains, but which nonetheless can be individu-
ally outweighed by a sufficient number of mild ankle sprains. Perhaps
a broken ankle is such a misfortune. Even though it is worse that one
person break her ankle than that she mildly sprain it, it is worse that
many people have mild ankle sprains than that one has a broken ankle.
But this process of escalation can be continued. For each misfortune
short of the worst possible one, there is a worse misfortune that can be
individually outweighed by a sufficient number of the lesser one. In par-
ticular, it seems plausible that there is some misfortune short of death,
perhaps some kind of mutilation, that can, if suffered by enough people,
outweigh one death. Consider now a sequence of judgments, S, that
begins as follows: one death is better than n' mutilations; n' mutilations
are better than n2 xs (where x is some misfortune less bad than mutila-
tion). S continues with the first term of each comparison being identical
to the second term of the previous comparison, until we reach the last
two comparisons: nm-2 broken ankles are better than nm-i mild ankle

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
sprains; nm-i mild ankle sprains are better than nm mild headaches. If we
have S, we can conclude, by the transitivity of 'better than' that one
death is better than nm mild headaches. In which case, we must reject
Less.
How might a consequentialist attempt to salvage Less? There are two
possibilities that I will discuss. First, she might deny the existence of any
true S. Perhaps there are some misfortunes so bad that even one such
misfortune is worse than any number of any lesser misfortunes. Death
is the most likely candidate for such a super misfortune. There is some-
thing special about death. As Clint Eastwood says in Unforgiven, "It
takes away all a man has, and all he's gonna have." Unpleasant as even
severe mutilation is, it is still worse that one person die than that any
number are mutilated. In which case, the first comparison in S will be
false. This view of death may appeal to those who believe what students
in introductory classes say when they claim that life is 'invaluable' or
'infinitely valuable,' but is it really plausible? Can anyone who really
considers the matter seriously, especially a consequentialist, honestly
claim to believe that it is worse that one person die than that the entire
sentient population of the universe be severely mutilated? My concern
here is not to persuade someone who claims that even one death is
worse than any number of any lesser misfortunes. I merely wish to ex-
plain what is involved in rejecting S. Perhaps the break in the sequence
of misfortunes could occur at some later point. Perhaps there is some
misfortune, short of death, that is worse than any number of any lesser
misfortunes. This seems even more implausible, though, than the claim
that death is worse than any number of any lesser misfortunes.
III
Even if we accept S, we may yet attempt to hang on to Less. Recall that

the rejection of Less followed not from S alone, but from S and the tran-
sitivity of 'better than.' If we reject the transitivity of 'better than,' we
may accept both S and Less. Consider a simpler sequence, S': one death
is better than a million broken legs; a million broken legs are better than
ten billion mild headaches. If we accept both S' and the transitivity of
'better than,' we must accept that one death is better than ten billion
mild headaches, and thus reject Less. But if we reject the transitivity of
'better than,' we may accept both S' and Less.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
Not only does the claim that 'better than' is intransitive allow a conse-
quentialist to accept both Less and the highly plausible S, but it also
provides a way of accommodating moral dilemmas within a consequen-
tialist framework. Suppose we accept both S' and the claim that ten bil-
lion mild headaches are better than one death. Consider the following
scenario:
Intransitive Trolley: A three-car trolley is speeding out of control to-

ward a three-way fork in the track. The driver, who is a maximizing
consequentialist, cannot stop the trolley. If he does nothing, one car
will hurtle down each track. He does, however, have three other op-
tions, corresponding to three buttons on his control panel. If he
pushes button 'L,' all three cars will go down the left track. If he
pushes 'M,' all three cars will go down the middle track. If he pushes
'R,' all three cars will go down the right track. If one or more cars enter
the left fork in the next minute, exactly one innocent person will be
killed. If one or more cars enter the middle fork in the next minute,
exactly one million innocent people will suffer broken legs. If one or
more cars enter the right fork in the next minute, exactly ten billion
innocent people will suffer mild headaches.
What should the driver do? One of his options is clearly worse than all
the others. Doing nothing is worse than pushing any of the buttons.
However, there is no best option, no option that is at least as good as all
other options, and no option than which there are none better. Of the
three button-pushing options, each one is better than one other and
worse than one other. L is better than M, but worse than R. M is better
than R, but worse than L. R is better than L, but worse than M. There is
no permissible option. The driver is thus faced with a moral dilemma.
Notice that it does not matter which version of the permissibility re-
quirement we employ, Con i or Con 2. In this sense, the denial of the
transitivity of 'better than' is a more effective way of accommodating
moral dilemmas within a consequentialist framework than the claim
that some values (or options) are incomparable.
Should a consequentialist embrace intransitivity? The considerations
of the previous paragraph will probably not sway many consequential-
ists, who tend to reject the claim that there are moral dilemmas5 (as
opposed to moral conflicts, in which even the best option is regrettable).
5. See, for example, Alastair Norcross, "Should Utilitarianism Accommodate Moral Di-
lemmas?" Philosophical Studies 79, no. 1 (July 1995): 59-85.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
Is Less such a compelling principle that we should reject something as

overwhelmingly plausible as the transitivity of 'better than' in order to
accommodate it? There are, in fact, considerations that directly under-
mine Less. For example, it is reasonable to suppose that, at least some-
times, raising the speed limit on highways is better than keeping it at its
current level. The benefits of increased speed limits, however, are
merely increased convenience for many, while the losses are the deaths
of a few. I will postpone a fuller consideration of this example until later
in this article. I mention it here simply to raise the possibility that the
attractiveness of Less may be no more than skin deep. The transitivity
of 'better than,' on the other hand, is deeply entrenched. In fact, some
may even regard it as an analytic truth. Clearly, then, additional argu-
ment is needed to convince a consequentialist to embrace intransitivity.
IV
In this and the next section, I will consider two kinds of attempt to argue
for the intransitivity of 'better than.' First I will examine a sorites-type
argument, presented by Warren Quinn, that uses a sequence of imper-
ceptible differences in utility that add up to a vast change. Second, I will
examine an argument of Larry Temkin's, originally devised by Stuart
Rachels, that our experiences of pain form a continuum, such that a
sequence of definite changes for the worse can add up to a change for
the better.
Warren Quinn, in "The Puzzle of the Self-Torturer,"6 presents the ex-
ample of a self-torturer hooked up to a device "that enables doctors to
apply electric current to the body in increments so tiny that the patient
cannot feel them. The device has iooi settings: o (off) and i ... 1ooo." At
each setting, the self-torturer is offered $10,000 to advance to the next
one. Since he can feel no difference between adjacent settings, he al-
ways prefers to advance a setting, and receive the money, rather than to
stay put. However, by the time he reaches iooo, he is in so much pain
that he prefers to return to o and relinquish all the money.
Quinn uses the example to argue, not that 'better than' is intransitive,
but that the torturer's stepwise preferences are intransitive. However, if
Quinn is correct in his claim that the torturer really is just as comfortable
at each setting as at the previous one, his example provides a conse-
6. Philosophical Studies 59 (1990): 79-90.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
quentialist, such as a utilitarian, with a reason to reject the transitivity

of 'better than.' Consider any choice between adjacent settings, S and
S+i. If the torturer chooses S+i, he will not be any less comfortable than
if he stays at S. Furthermore, he will get something he cares about,
$10,000, that he will not get if he stays at S. There are no respects in
which he is worse off at S+i than at S, and there is at least one respect
in which he is better off. The self-torturer will, therefore, be better off
all-things-considered at S+i than at S. If we consider two worlds, W and
W+i, that differ only in that the self-torturer is at S in W and at S+i in
W+i, we should judge that W+i is better than W However, according to
Quinn, the self-torturer is worse off at iooo than at o. So the world in
which he is at iooo is worse than the one that differs from it only in that
he is at o. Thus, intransitivity looms.
The argument of the previous paragraph depends crucially on the
claim that the torturer is just as comfortable at each step as at the previ-
ous one. Why should we accept this? Quinn considers several reasons
to reject this, and dismisses them, rather too hastily. For the sake of
brevity, I will consider just two of them. We might point out that Quinn's
claim is only plausible while we focus on the torturer's introspective
ability to distinguish between his comfort level at adjacent settings.
rhere may, however, be behavioral evidence, unavailable to introspec-
tion, that the torturer's comfort level is declining- "he might look less
comfortable, or be grouchier." Quinn's response to this possibility is
simply to stipulate that he is interested in a case in which "the individual
increments of current are too small even to have these effects." This
hardly gives the objection a run for its money. If we consider all the
behavioral evidence there might be, it is not clear how the increments
of current could be so small that no increase made any difference over
the previous setting. At o the torturer is not screaming, even a little bit.
At 1ooo he is screaming a lot. If his behavior is no different at i than at
o, he is not screaming, even a little bit, at i. Likewise, he is not screaming,
even a little bit, at 2. And so on. So he is not screaming, even a little bit,
at iooo. But he is. Clearly, there must be objective differences that are
made by some of the increments of current. To say that there is an ob-
jective difference is not necessarily to say that the torturer must move
from determinately not screaming to determinately screaming. Perhaps
at one setting he is determinately not screaming. At the next, it is no
longer determinately true that he is not screaming. This would still con-

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
stitute an objective difference. If so, it is possible that his comfort level

really does decline when he moves from one setting to the next, no mat-
ter how small the increment. In which case we do not have to accept
that he is really better off at each setting than at the previous one.
Quinn also considers the following criticism:
We are ignoring the effects of "triangulation": The self-torturer can

triangulate a difference between s and s+i if he can find some third
setting s' that feels the same as s but better than s+i. And if he can use
o to triangulate such a difference, then it is obvious that his comfort,
at least compared to o, declines in stepping from s to s+i.
His response is to stipulate, again, that he is interested in a case in which

the individual increments of current are "too small to make any differ-
ence in comfort, even one that can be detected only by triangulation."
This is puzzling. After all, no matter how small the increments of cur-
rent, the self-torturer eventually reaches some settings that feel worse
than o. At the very least, o feels better than the last setting. It is also
stipulated that o feels the same as i. If he simply considers each setting
in turn, why will there not be a first one that feels worse than o, or at
least a first setting of which it is no longer determinately true that it feels
the same as o? Quinn responds as follows:
But, empirically speaking, his preferences as between s and o can ex-

hibit various kinds of indeterminacy. Not only is there no empirically
determinable first setting that he disprefers to o, there is no empiri-
cally determinable first setting at which these preferences become in-
determinate.
Imagine the self-torturer comparing successive settings with o. In front

of him is a control panel with three buttons, labeled 'Yes,' 'No,' and 'It's
Indeterminate.' After performing each comparison, he is asked, "Does
this setting feel the same as o? Push the appropriate button within the
next sixty seconds." If he responds within one minute, the message cor-
responding to the button pushed appears on a screen. Otherwise no
message appears. If this procedure is performed, there will clearly be a
first setting at which 'yes' does not appear on the screen.7
7. For similar, steplike, arguments against vague objects, see Mark Heller, "Against
Metphysical Vagueness," Philosophical Perspectives lo, Metaphysics (1996), and Roy So-
rensen, "A Thousand Clones," Mind 103 (1994).

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
But perhaps I am being unfair to the case. I have only given the self-
torturer three options (four, if you include not pressing any button).
Why assume that the only possibilities for any setting are that it feels the
same as o, it does not feel the same as o, and it is indeterminate whether
it feels the same as o? Perhaps there are settings for which it is indeter-
minate whether it is indeterminate whether it feels the same as o. Call
such a state of affairs '2indeterminacy.' But then, will there not be a first
setting of which it is 2indeterminate whether it feels the same as o? Not
necessarily. Perhaps there are settings of which it is indeterminate
whether it is 2indeterminate whether it feels the same as o. Call such a
state of affairs '3indeterminacy.' This suggests that any claim to the ef-
fect that there must be a first setting of which it is nindeterminate
whether it feels the same as o can be countered with n+iindeterrninacy.
This move will work for any n.
However, now let us say that it is superindeterminate whether a set-
ting feels the same as o, just in case there is any indeterminacy of any
kind or at any level about whether the setting feels the same as o. Can
we now block the claim that there will be a first setting of which it is
superindeterminate whether it feels the same as o? Can there be
settings of which it is indeterminate whether it is superindeterminate
whether they feel the same as o? Clearly not. The postulation of indeter-
minacy about superindeterminacy is self-defeating. So now it appears
that there will be a first setting, of which it is either superindeterminate
whether it feels the same as o, or false that it feels the same o. In moving
from the previous setting to this setting, there is an objective change in
the torturer's comfort.
I conclude that Quinn has not succeeded in showing that 'just as
comfortable' is intransitive, and thus that he has not provided a conse-
quentialist with a reason to reject the transitivity of 'better than.'
Larry Temkin also argues that 'better than' is intransitive; his argument
centers on a variation of an example of Stuart Rachels,8 and rests on the
following three claims.
8. Larry S. Temkin, "A Continuum Argument for Intransitivity," Philosophy & Public
Affairs 25, no. 3 (Summer 1996): 175-210. All page references in the text are to this article,
unless otherwise stated. Temkin offers a different argument for intransitivity, based on

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
Claim 1: for any unpleasant or "negative" experience, no matter what

the intensity and duration of that experience, it would be better to
have that experience than one that was only a little less intense but
twice as long.
Claim 2: there is a continuum of unpleasant or "negative" experiences
ranging in intensity, for example, from extreme forms of torture to the
mild discomfort of a hangnail.
Claim 3: a mild discomfort for the duration of one's life would be

preferable to two years of excruciating torture, no matter the length
of one's life. (179)
Claims 1-3 all "depend on an appropriate 'other things equal' clause"

(182). Now, we are to consider a sequence of comparisons of lives, all
overall worth living, A with B, B with C, C with D .... X with Y, that
begins as follows: "First, compare two lives A and B. Suppose that A and
B are both lengthy-perhaps, indeed, verylengthy-and that A and B are
similar, except that A contains two years of excruciating torture, B four
years of torture whose irntensity is almost, but not quite, as bad as
A's.... Next, compare B with C, where C stands to B as B stands to A"
(180). The sequence continues, with each successive life containing a
pain that is a little less intense but twice as long as the pain in the previ-
ous life. At the end of the sequence, "one would be comparing two alter-
natives, say, X and Y, such that X had an annoying hangnail for a very
long time-perhaps thousands of years9-and Y had a hangnail that was
almost, though not quite, as unpleasant as X's, but that lasted twice as
long" (180). In accordance with claim i, A is better than B, B is better
than C .. ., X is better than Y. If 'better than' is transitive, A is better than
Y. But claim 3, and the intuitions of many when first presented with the
case, tell us that A is not better than Y. Thus transitivity is threatened.
what he calls the 'Person-Affecting Principle,' in "Intransitivity and the Mere Addition Par-
adox," Philosophy & Public Affairs 16, no. 2 (Spring 1987): 138-87. I discuss this argument
in my "Intransitivity and the Person-Affecting Principle," American Philosophical Associ-
ation annual meeting, Central Division in Pittsburgh, May 1997. Rachels's version of the
example appears in "Reconceiving 'Better Than,"' American Philosophical Association
annual meeting, Pacific Division in San Francisco, April 1995, and "Counterexamples to
the Transitivity of 'Better Than,"' Australasian Journal of Philosophy, forthcoming.
9. Actually, if the sequence contains lives corresponding to all the letters A through Y,
X's hangnail would last for more than 16 million years, and Y's for more than 32 million
years.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
This argument is provocative, and initially c

main unconvinced. Claims i and 2 are relativel
as they are interpreted loosely with the relevant 'other things being
equal' clauses.10 It is claim 3 that I find problematic. First, let us be clear
about what it says in the context of this example. Claim 3, as it applies
to the comparison of A with Y, tells us that a life that is 32 million years
long, overall worth living, and that contains the discomfort of a mild
hangnail for the entire 32 million years is preferable to a life that is also
overall worth living, and is just like it," except that, for two of the 32
million years it contains excruciating torture instead of hangnail pain,
and for 31,999,998 years it contains neither torture nor hangnail pain.
Both claim 3 and the judgment that A is worse than Y are initially
intuitively appealing. But how trustworthy are our intuitions, when it
comes to a comparison between A and Y? How many of us have had the
misfortune to experience excruciating torture at all, let alone for two
years? I have experienced pains no more severe than a broken wrist, torn
ankle ligaments, or an abscess in a tooth. These were pretty bad, but I
have no doubt that a skilled torturer could make me experience pains
many times as bad. I have experienced a mildly annoying hangnail.
When I try to imagine two years of pain that is many times worse than
the worst pain I have ever experienced, and I consider whether any
amount of hangnail pain could outweigh it, the immediate answer is no.
But just what am I considering? Two years is a fairly sizable portion of
a human life. How could any life with that large a portion spent in uni-
maginable pain be better than a life with only a mildly annoying hang-
10. Thus, we would not insist on an infinitely divisible continuum for claim 2, or on the
less intense experience of claim 1 being twice as long, as opposed to even longer. Temkin
makes these points himself:
More to the point, for any imaginable unpleasant experience, we can imagine another
that would be noticeably better, but still sufficiently bad that we would prefer the former
to twice (or three or five times) as much of the latter. This suggests that even if, phenom-
enologically, there isn't a smooth continuum, there is still a discontinuous spectrum of
possible experiences ranging from A to Y that is sufficient for my purposes. (183-84)
11. The two years of torture have no effect on the rest of A. There are no painful memo-
ries afterwards, and no dreaded anticipation beforehand. Likewise, the hangnail pain re-
mains constant throughout Y. It does not "eventually have the effect of the so-called Chi-
nese water torture-where a steady drip of water ultimately drives one crazy" (181). But
neither does one become used to it, so that eventually an extra day, or year, or even million
years makes no difference. If that were the case, it would not be clear that Y was worse
than X.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
nail throughout? But in comparing A with Y we are supposed to imagine

lives many times longer than any person has ever lived. I suggest that
our intuitions about the comparison of A with Y are unreliable in at least
two respects. First, it could be that our intuitions overestimate the sig-
nificance of the two years of torture in A, because we simply cannot
imagine what it would be like to live for 32 million years. Two years may
be a sizable portion of a standard human life, but it is a tiny fraction of
32 million years. Proportionally, it is the equivalent of 2.6 minutes of
torture in an eighty-year human lifetime. (Would A seem worse than Y,
if the torture were to be experienced for ten seconds every five years?
Clearly not.) Second, we underestimate the significance of the 32 million
years of hangnail pain in Y. I have trouble imagining what it would be
like to have a hangnail for even a few years, let alone an entire ordinary
lifetime. And yet, when I try to compare A with Y, I am supposed to
imagine having a hangnail continuously for 32 million years!
Temkin considers at least the second criticism of the previous para-
graph. His reply, which applies equally to the first criticism, is twofold.
First, he claims that, evefi if claim 3 is false, his argument will still work,
because A and Y will not be anything like as lengthy as I suggest. Second,
he defends claim 3 directly. I will examine his first reply first.
Temkin claims that we may be able to traverse the pain spectrum
from excruciating torture to mildly annoying hangnail in a relatively
small number of steps, each of which is small enough to fit the original
structure of the argument. Perhaps, he suggests, our experience of pain
is analogous to our experience of color:
Consider the following spectrum: red, reddish orange, orange, orang-

ish yellow, yellow, yellowish green, green, greenish blue, blue, indigo-
blue, indigo, indigo-violet, violet. It is hard to deny that, phenomenol-
ogically, the gaps between the adjacent members of this spectrum are
pretty small, though still clearly perceptible. Yet it would only take
twelve steps to get from one extreme of the spectrum to the other.
(186)
If we could get from excruciating torture to mildly annoying hangnail in

twelve steps that are small enough to fit the structure of the original
example, A and Y (or whatever the last life in the sequence would be
called) would be a 'mere' 8192 years long. That is certainly less of a
stretch for our human imaginations than 32 million years, but I am still

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
not ready to trust my (or Temkin's) intuitions on the comparison of A

with Y. But, let us grant, for the sake of argument, that we can confi-
dently proclaim that 8192 years of a hangnail is better than two years of
torture and 8190 years of neither torture nor hangnail (other things
being equal). Why should Temkin's color comparison persuade us that
A and Y could be as short as 8192 years? Temkin concedes that the color
analogy is "purely suggestive," but he adds that he "would be surprised
if our phenomenological powers of discrimination differed radically
across our sense modalities" (188).
But what would it be for the case of pain to be analogous to that of
color, in the way required for Temkin's argument? Say we divide the
distance from torture to hangnail into twelve equal steps.'2 Likewise, we
have twelve steps from red to violet.'3 Now, the crucial question is how
small are the steps across the pain spectrum? Well, we say, if they are
analogous to the steps across the color spectrum, they are pretty small.
After all, the step from, say, yellow to yellowish green is pretty small. But,
this is too vague to be any help. We need to know whether the steps
across the pain spectrum are small enough for the purposes of Temkin's
argument. That is, it must be clear that x years of one pain is preferable
to 2X years of the pain that is just one step down from it. How do we
pursue the analogy with color? Is one year of yellow clearly better than
two years of yellowish green? Or perhaps we should ask whether one
year of yellowish green is clearly better than two years of yellow. Neither
question makes sense. This is precisely the problem with arguing, even
suggestively, from the case of color to the case of pain. Comparisons of
pains have natural evaluative import. If one pain is greater than another,
it is, other things being equal, preferable to experience the latter. The
pain spectrum has a natural evaluative direction, from greater pains to
lesser pains. There is nothing analogous with color comparisons or the
color spectrum. There is no natural sense in which red is preferable to
or greater than orange, or vice versa.
Perhaps we could try to force evaluative significance on the color
spectrum with a thought experiment. Imagine beings whose qualitative
12. The steps must be equal in the proportionate sense. In the sense in which the step
from 5 to 10 is equal to the step from io to 20, not to the step from io to 15. Call the sense
in which the step from 5 to 10 is equal to the step from 10 to 15 additive equality.
13. Here we have the first problem with the analogy. It is not clear what, in the color
case, would correspond to the difference between proportionate and additive equality in
the pain case. If anything, Temkin's twelve steps across the color spectrum seem to be
equal in the additive sense, but he needs them to be equal in the proportionate sense.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
states are just like ours, with the following addition. The experience of
color is painful for them. The pain of experiencing red is the same inten-
sity as excruciating torture. The intensity diminishes through the color
spectrum, with the pain of experiencing violet having only the intensity
of a mildly annoying hangnail. The qualitative feel of a color experience
is the same as ours, with the addition of the pain. Imagine now the
difference between the experience, for such a being, of yellow and of
yellowish green. Since part of those experiences is the same as ours, the
nonpainful part that is, we have a perfectly clear idea of how similar they
are in that respect. But the respect we are interested in is how much
better is the experience of yellowish green than the experience of yellow.
In particular, how long a yellowish green experience would it take to
outweigh, say, one hour of yellow experience. To answer that question,
we need to know how the pain parts of the experiences compare. So we
need to divide the pain spectrum, from torture to hangnail, into twelve
equal (proportionate) steps, and ask how much of one pain would it take
to outweigh, say, one hour of the next pain up. But this is the question
the color analogy was supposed to help us with. Instead, it seems that
we cannnot make sense of the color analogy without first answering this
question.
I have been arguing that Temkin's color analogy gives us no reason
to believe that we could traverse the pain spectrum in as few as twelve
steps, all of which are small enough to keep lives A and Y to (roughly)
imaginable lengths.'4 However, perhaps we need not settle the question
of how many (small enough) steps it would take for us to get from excru-
ciating pain to a mildly annoying hangnail. Perhaps, as a matter of em-
pirical fact, it would take us so many steps to get from one end of the
14. Temkin also tries a variation of his example, that envisages a cardinal ranking of
pains from 1 (mildly annoying hangnail) to loo (excruciating torture). On this scale, a
moderately uncomfortable limp is ii. If we start with two days of excruciating torture, and
move in steps that decrease the intensity of the pain by 20 percent but double its duration,
we reach 5.6 years of a moderately uncomfortable limp in ten steps. Temkin claims that,
other things being equal (the torture leaves no memories, the limp has no side-effects,
etc.), two days of torture is worse than 5.6 years of the limp. Even if I agreed with that
judgment, I would find the argument unconvincing. On Temkin's scale, excruciating tor-
ture is fractionally more than nine times worse than a moderately uncomfortable limp. I
have had a moderately uncomfortable limp, when I was recovering from torn ankle liga-
ments. In my judgment, the pain of the freshly torn ligaments was far more than nine
times as bad as the limp. Furthermore, as I stated above, I am quite sure that an experi-
enced torturer could make me experience pain many times worse than the pain of freshly
torn ligaments.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
pain spectrum to the other that A and Y would be millions (or even
billions) of years long. Temkin's response is that, even if this is true, it
is surely only contingently true. All he needs is the bare possibility of
beings whose experience of pain is different enough that A and Y, for
them, would be only hundreds, or perhaps thousands, of years long.'5
Imagine beings whose experience of pain is sufficient for Temkin's pur-
poses. Call them 'intransitives.' If it is metaphysically possible for there
to be intransitives, the axiom of transitivity for 'all things considered
better than' is false. Even if we are reasonably sure that we are not in-
transitives, surely it is (metaphysically) possible for there to be intransi-
tives. So, the axiom of transitivity is false.
This argument is simple and appealing, but we must be careful not to
slide from epistemic possibility to metaphysical possibility. Suppose, for
the sake of argument, that we just do not know whether there are intran-
sitives. For all we know, there are intransitives. That is, it is epistemically
possible that there are intransitives. Or perhaps we are fairly sure that
there are no intransitives, but, for all we know, there may be. So, it is
epistemically possible that there may be intransitives. Nothing follows
from either claim of epistemic possibility about the metaphysical possi-
bility of intransitives. (Imagine an agnostic, who holds that, for all he
knows, God exists, and, for all he knows, God does not exist, and who
believes that God, if there is one, is a necessary being. If he slides from
the epistemic to the metaphysical possibility of God's existence and
nonexistence, he will be committed both to the view that God exists and
the view that God does not exist!)
Why should we believe that it is metaphysically possible for there to
be intransitives? Here is a bad argument, which might underlie an intu-
ition that intransitives are metaphysically possible. It is clearly a contin-
gent matter that we experience pain the way we do. We could have been
constructed differently, and have had different experiences of the pain
spectrum. So, we could have had the experiences of the intransitives.
But, even if it is true that our experiences of pain could have been differ-
ent, it does not follow that they could have been different in just any
way. The main quad of Christ Church, Oxford, could have had a differ-
15. "Unless it is metaphysically impossible for the pain and color modalities to be anal-
ogous in the way suggested, there are some possible creatures for whom my argument
would work-even if it does not, in fact, work for us. This is all one needs to undermine
the claim that 'all things considered better than' must be transitive" (188).

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
ent shape and/or dimensions. But here is one combination of shape and
dimensions it could not have had: square, with sides of 200 feet each
and a diagonal of loo feet.
Here's another bad argument that, until exposed, could be driving our
intuitions. Even if our pain spectrum spans such a large range from
excruciating torture to mildly annoying hangnail, we might have been
constructed differently, so that the range from the worst pain we could
experience to the least pain was much smaller. In particular, we might
have been able to get from the worst pain to the least pain in a small
number of steps, say, twelve, each of which was small enough that a
certain duration of one pain was clearly preferable to twice the duration
of the next pain down. Call this possible feature of pain experience 'Q.'
Since having Q is all it takes to be an intransitive, there could have been
intransitives. The problem is that Q is not all it takes to be an intransi-
tive. Our experience of pain needs to have Q and the following further
feature, call it 'I': 8192 years of the least pain is clearly better than two
years of the worst pain and 8190 years of neither the worst pain nor the
least pain (other things being equal). I would seem to require a pain
spectrum as broad (roughly) as ours. Many possible creatures could
have Q but lack I. Consider a creature whose pain spectrum ranges from
the pain of a broken leg to the pain of a sprained ankle. That is, it ranges
from the pain that a typical human suffers from a broken leg to the pain
that a typical human suffers from a sprained ankle. Such a creature
would probably have Q. They could get from their worst pain-the pain
of a broken leg-to their least pain-the pain of a sprained ankle-in
twelve small enough steps. But would 8192 years of sprained ankle pain
be clearly better than two years of broken leg pain and 8190 years of
neither sprained ankle pain nor broken leg pain (other things being
equal)? Clearly not. Temkin's argument depends on the ends of the pain
spectrum being roughly as far apart as excruciating torture and a mildly
annoying hangnail. While it may be fairly easy to imagine beings whose
pain spectrum is narrow enough that they could get from one end to the
other in twelve small enough steps, it is not so easy to imagine beings
who could span a spectrum that is roughly as broad as ours in twelve
such steps.
Let me illustrate what I take to be the problem with the claim that
intransitives are metaphysically possible. My back yard is about 120 feet
long. My stride is about 3 feet. So I can traverse the length of my back

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
yard in 40 strides. Could it have been the case that some creature could
have traversed my back yard in lo strides? Yes. Perhaps there could have
been a creature with a 12-foot stride, or perhaps my back yard could
have been only 30 feet long. But, could there have been a creature with
the same length stride that I actually have, who could have traversed a
backyard of the same length as my actual back yard in only lo strides?
Clearly not. When we ask whether there could be intransitives, we are
asking whether there could be creatures, some of whose experiences
violate the axiom of transitivity for 'all things considered better than.' If
we believe, as I do, that 'all things considered better than' is transitive
for all possible alternatives, we have a strong reason to reject the meta-
physical possibility of intransitives. If we reject the axiom of transitivity
for 'all things considered better than,' we lack at least that reason for
rejecting the metaphysical possibility of intransitives. I do not see how
we can consider the metaphysical possibility of intransitives, in isola-
tion from our belief in the axiom of transitivity, in an attempt to shape
that belief.
I conclude that we have no good reason to suppose that any possible
pain experiences could span the pain spectrum from torture to hangnail
in few enough small enough steps that A and Y were 'merely' a few
hundred (or even thousand) years long. However, it is still possible to
accept Temkin's argument, if we are prepared to accept claim 3, despite
the problem of trusting our intuitions regarding pains of unimaginably
long duration. So, I turn now to a direct consideration of claim 3.
Perhaps we think that two years of excruciating torture is so bad that
only more of the same kind of suffering could be worse than it. And we
think this, even if the two years of torture is a tiny fraction of an
enormously long life, the rest of which is happy enough (perhaps bliss-
fully happy) to make the life overall worth living. B is worse than A,
because B contains the same kind of suffering as A, albeit a little less
intense, and lots more of it. The suffering in Y, on the other hand, even
though almost unimaginably lengthier than the suffering in A, is clearly
not of the same kind. So Y cannot be worse than A.16 A little thought will
tell us that this line of reasoning will not justify the claim that A is worse
than Y. Recall that A is overall worth living. It is, therefore, better than
any life that is not overall worth living. Is it not possible to alter Y in such
16. Temkin seems to endorse this explanation on p. 194.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
a way that it is not overall worth living, without introducing the same
kind of suffering as the torture in A? Clearly, a life of unmitigated excru-
ciating torture would not be worth living. But there may be many lives
that are not worth living, that contain nothing like excruciating torture
at any point. Imagine 32 million years (or even just fifty or so years) of
tasteless food, unfulfilling relationships, ill-fitting, drab, uncomfortable
clothes, a tedious job with contemptuous and contemptible coworkers,
a drafty, dirty one-room apartment with a lumpy bed, a thoroughly
uninquiring intellect, and a continuous mildly annoying hangnail. Call
such a life "YME." It seems highly likely that YME would not be worth
living, and so would be worse than A. (If you think that YME would be
worth living, just keep adding details until you get to a life you consider
not worth living. I think you will find you can do it, without introducing
anything like excruciating torture.) Yet YME contains none of the same
kind of suffering as A. So, two years of excruciating torture can be out-
weighed by enough fairly minor suffering and deprivation. Why, then,
can it not be outweighed by an enormously lengthy hangnail pain? My
example, of course, does not prove that two years of excruciating torture
can be outweighed by some amount of hangnail pain, but it does under-
mine a principle which may well be the basis for our intuitive assent
both to claim 3 and to the specific judgment that A is worse than Y.17
Let us look now at Temkin's explanation of why two years of torture
cannot be outweighed by any amount of hangnail pain:
[I]n essence, I think significant amounts of torture have lexical prior-

ity over any amount of a hangnail, just as I think lives have lexical
priority over lollipop licks. My model for this is something like the
following. Torture's badness might range from o to lo, depending on
its duration, with two years of torture being, say, a 7. A hangnail's
badness might range from o to 1. Prolonging a hangnail increases the
value of the decimal places representing its "badness score," but the
fundamental gap between 1 and 7 is never affected. (191)
What do the numbers signify in this account? The claim that two years
of torture is a 7, on a scale of o to lo, does not tell us that any life contain-
17. Perhaps Temkin could respond to my example of YME by altering his example, so
that all the lives A ... Y are not worth living. Rachels's version of the example (the most
recent version that I have seen) has this structure. Such a move has other difficulties,
though, that I do not have the space to explore here.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
ing two years of torture is equally bad, nor does it tell us that four years
of torture is not, other things being equal, twice as bad as two years of
torture. It tells us that any life containing two years of torture is worse
than any life of equal length, which is just like it, except that instead of
torture it contains some amount of pain whose badness ranges from o
to less than 7. Since Y is of equal length with A, and just like it except
that, instead of two years of torture, Y contains a certain (very large)
amount of a pain, whose badness ranges from o to 1, A must be worse
than Y.
Presumably, there are some pains worse than a hangnail whose bad-
ness ranges from o to less than 7. A will be worse than any life in the
sequence that contains such a pain. X, for example, contains about 16
million years of annoying (as opposed to mildly annoying) hangnail
pain. I should think Temkin would want to say that the badness of an-
noying hangnail pain ranges from o to considerably less than 7. Thus,
A must be worse than X. There must also be some pains less intense than
excruciating torture, whose badness ranges from o to more than 7. The
slightly less excruciating torture in B would be such a pain. Since we
know that B is worse than A, four years of the pain in B must be more
than 7, even if not by much. Let us say, then, that four years of the pain
in B has a badness score of 7+x. Now, since we also know that C is worse
than B, eight years of the pain in C must have a badness score of more
than 7+x, even if not by much. Let us say, then, that eight years of the
pain in C has a badness score of 7+x+y. The problem now is obvious.
How are we ever going to get to a pain with a badness score of less than
7? Since each life in the sequence is worse than the previous one, the
badness score for the specific duration and type of pain in each life must
be higher than that for the corresponding pain in the previous life. Thus,
either the range for a mildly annoying hangnail is from o to more than
7, which contradicts this model for understanding claim 3, or at least
one life in the sequence is clearly better than the previous one, which
contradicts claim 1.
Perhaps Temkin's numerical model for explicating claim 3 is merely
infelicitous. Let us try to express the same idea without recourse to
numbers. A is worse than Y, because any life containing two years of
excruciating torture is worse than any life of equal length that is just like
it, except that, instead of the torture, it contains some (perhaps very
lengthy) duration of mildly annoying hangnail pain. Let us express this

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
as the claim that A has the property of being worse than any amount of
hangnail pain.'8 In order for this to be genuinely explanatory, we are to
think of this as an intrinsic feature of A, which has relational conse-
quences. It is like the claim that Tom is over six feet tall, rather than the
claim that Tom is taller than Dick, who happens to be five feet tall, and
Harry, who happens to be four feet tall.'9 Call all and only lives that have
the property of being worse than any amount of hangnail pain 'misera-
ble.' Is B miserable? I should imagine that Temkin would claim that it
is. X, on the other hand, clearly is not miserable. It cannot be, since, by
claim 1, X is better than Y. Consider now the following claim:
Claim 4: between any two lives, x and y, of equal length, that differ
only in that x has a certain duration of one pain and y has a certain
duration of a pain of different intensity, if x is miserable, then x is
better than y only if y is also miserable.
Claim 4 is, on inspection, just as compelling as claims 1-3. (In fact, most
would assent to a stronger claim, according to which y is better than x,
if y is not miserable.) But -now we have trouble reconciling claims 1, 3,
and 4. Consider the sequence A ... Y. According to claim 3, A is misera-
ble. Clearly Y is not miserable. So, there must be a first life, a, in the
sequence that is not miserable. By claim 4, then, the previous life in the
sequence is not better than a. But this contradicts claim i. Perhaps inde-
terminacy will rear its ugly head again. Maybe there are no adjacent lives
in the sequence such that the first determinately is miserable, and the
second determinately is not. However, there will be a first life, ,B, of
which it is no longer determinately true that it is miserable. According
to claim 4, then, it is not determinately true that the previous life in the
sequence is better than ,. This still contradicts claim 1. Perhaps Temkin
could come up with some variations on claims 1 and 3 that are compel-
18. A life is worse than any amount of hangnail pain, just in case it contains some
duration of some pain such that, any life containing that duration of that pain is worse
than any life of equal length that is just like it, except that, instead of that pain, it contains
some duration of mildly annoying hangnail pain.
19. Suppose that having the property of being worse than any amount of hangnail pain
were a purely relational feature of A. In that case, the explanation that A is worse than Y
because A is worse than any amount of hangnail pain has the same explanatory force as
the claim that A is worse than Y because A is worse than X or Y or Z. Another reason to
construe the property of being worse than any amount of hangnail pain as intrinsic, is so
that claim 4, which employs this property, does not have the appearance of presupposing
transitivity.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
ling, can coexist happily with claim 4, and w

However, I suspect that I could come up with
would once again threaten the harmony.
The previous two arguments share a common structure. If Temkin is
to explain why A is worse than Y, rather than simply assert that it is, he
must appeal to a property possessed by A but not by Y (e.g., having a
badness score of more than 1, being miserable). Given the 'other things
being equal' clause that applies to all the lives in the sequence, the pos-
session by A but not by Y of the property is sufficient for the judgment
that A is worse than Y. There will, however, be a first life in the sequence
that does not possess the property. Therefore, there will be an adjacent
pair of lives, say m and n, such that m does but n does not possess the
property. Therefore, n is not worse than m. But this contradicts claim
1.
I will close this section with a direct challenge to claim 3. First, let me
repeat the claim.
Claim 3: a mild discomfort for the duration of one's life would be

preferable to two years of excruciating torture, no matter the length
of one's life.
Imagine a life that is 500 years long and overall worth living, but that
contains two years of excruciating torture. (If you think that no life of a
mere 500 years which contained two years of torture could be worth
living, simply make it longer and adjust the rest of the numbers in this
example accordingly.) The 498 years not spent in torture would be toler-
ably good, certainly good enough to make the life overall worth living.
Call the condition during the 498 years 'bliss.' A life of two years of tor-
ture and 498 years of bliss is, therefore, worth living. What if we add 500
more years of bliss? Would such a life-2 years of torture and 998 years
of bliss-not be better than the original life? Perhaps such a life, call it
'A,' would even be well worth living. According to claim 3, though, iooo
years of hangnail pain would be better than A. (That is, iooo years, all
of which was just like bliss with the addition of a mildly annoying hang-
nail.) Claim 3, with the appropriate 'other things being equal' clause,
does not, however, entail that all lives containing just hangnail pain are
better than A. What about a life that consisted entirely of one hour of
hangnail pain? Would it be better to live for only one hour, during the
whole of which you had a hangnail, or to live for iooo years, two years

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
in torture and 998 years in bliss? Given that A is worth living, it seems
overwhelmingly likely that some lives containing a hangnail through-
out, call them 'hangnail lives,' are not better than A. In particular, it
seems likely that there is some hangnail life, call it 'HL,' such that it is
rational to be indifferent between living A and living HL.
Suppose that HL is loo years. (If you think it would have to be much
shorter, simply adjust the numbers accordingly.) Now imagine Bart and
Lisa choosing in advance how their lives will go, from among a limited
set of options. Each is given the choice only between A and HL.20 Bart
chooses HL and Lisa chooses A. Since it is rational to be indifferent
between HL and A, neither lives a better life than the other. Suppose
now that they are both able to extend their lives in one of two different
ways. At the end of the initial segments of their lives, they are both given
the choice between living for a certain duration in bliss or living for a
longer duration with a hangnail. Living with a hangnail is worse, albeit
not by much, than living in bliss. Therefore, for any particular duration
of bliss, t, there is a longer duration, t+, such that it is rational to be
indifferent between living for t in bliss and living for t+ with a hangnail.
The increased length of t+ over t compensates for the slight decrease in
quality between the blissful life and the hangnail life (recall that the
hangnail life is worth living). In particular, there is a length of bliss, such
that it is rational to be indifferent between that length of bliss and that
length plus goo years of a hangnail. Suppose that length to be lo,ooo
years (again, nothing rests on the particular choice of numbers). It is,
therefore, rational to be indifferent between lo,ooo years of bliss, call it
'B,' and lo,goo years of a hangnail, call it 'H.'
Bart and Lisa are each offered the choice between B and H. Not only
are they both rationally indifferent between B and H, but the experience
of B for Lisa is just like the experience of B for Bart, and likewise for H.2'
Lisa chooses B and Bart chooses H. Lisa's total life now consists of A
followed by B, Bart's of HL followed by H. So, Lisa's total life consists of
20. Furthermore, A for Bart is just like A for Lisa, and likewise with HL. For more details
on this specification, see the following note.
21. This specification is important in order to construct a pair of lives to which claim 3
is supposed to apply. Bart's hangnail life is supposed to be just like Lisa's bliss, with the
addition of the hangnail pain. Since the choice between B and HL is the choice between
lo,ooo years of bliss and lo,goo years just like bliss with the addition of hangnail pain, if
Bart's experience of bliss would not have been like Lisa's, his hangnail experience would
not have been just like her bliss with the addition of hangnail pain.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
two years of torture and 10,998 years of bliss. Bart's is just like it, except
that, instead of the torture, he has 11,ooo years of hangnail pain. Given
that neither A nor HL was preferable to the other, and that neither B nor
H was preferable to the other (and that B, for Lisa, was the same as it
would have been for Bart), neither Lisa's nor Bart's total life is preferable
to the other. But claim 3 says that Lisa's life is worse than Bart's. So we
should reject claim 3.22
Given the overwhelming plausibility of the claim that 'all things con-
sidered better than' is transitive, and the serious doubts I have raised
about Temkin's argument, I conclude that he has provided an interest-
ing and challenging, but ultimately unconvincing argument for intransi-
tivity.
VI
We have seen that neither Quinn's sorites argument nor Rachels's and
Temkin's continuum argument can provide good reasons to reject the
transitivity of 'better than.' Unless some other argument surfaces, a con-
sequentialist who wishes to embrace the overwhelmingly plausible S
must reject Less:
Less: The state of affairs constituted by any number of fairly minor

headaches is less bad than even one premature death of an innocent
person.
But, for a consequentialist, the rejection of Less, coupled with the rejec-
tion of incomparability, leads to life for headaches:
22. It might be objected that we can construct a pair of clearly unequal composite lives
out of choices just like the ones in my example. Suppose that Bart and Lisa are each
indifferent between studying the violin and studying the clarinet in college. Bart chooses
the violin and Lisa chooses the clarinet. At the end of college, they are both presented with
the choice between continuing their studies at ConservatoryA or Conservatory B. They are
both (rationally) indifferent between the two conservatories. Bart chooses A and Lisa
chooses B. It does not follow that Bart's experience of studying the violin in college and
in Conservatory A is neither better nor worse than Lisa's of studying the clarinet in college
and in Conservatory B. Perhaps A and B are the two (equally) best places to study the
clarinet and the two (equally) worst places to study the violin. In which case, Lisa's expe-
rience is probably better than Bart's. This example, though, runs foul of the 'other things
being equal' clause in claim 3. In my example, Lisa's experience of bliss is just the same
as Bart's would have been, had he chosen it. (See the previous note.) In the current exam-
ple, Lisa's experience of studying the clarinet at B (one of the best places for that purpose)
is clearly not the same as Bart's experience of studying the violin at B (one of the worst
places for that purpose) would have been.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
Life for headaches: there is some finite number of headaches, such

that it is permissible to kill an innocent person to avoid them.
A reader of this article, until this point, could be forgiven for assuming
that they were reading an argument against consequentialism. How-
ever, in this section, I will argue that life for headaches is not, after all,
particularly unpalatable. In fact, I will claim, most of us, consequential-
ists and nonconsequentialists alike, accept at least some other claims
that do not differ significantly from life for headaches.
Thousands of people die in automobile accidents every year in the
United States. It is highly probable that the number of deaths is positively
correlated with the speed limits in force on highways, at least within a
certain range. One of the effects of raising speed limits is that there are
more accidents, resulting in more deaths and injuries. One of the effects
of lowering speed limits is that there are fewer accidents. Higher vehicle
safety standards also affect both the numbers of accidents and the sever-
ity of the injuries sustained when accidents do occur. Another effect of
raising speed limits is that more gasoline is consumed, which raises the
level of particulate pollution, which also leads to more deaths.23 Stricter
standards for fuel efficiency also affect the amount of gasoline con-
sumed. There are, then, many different measures that we, as a society,
could take to lower the number of automobile-related deaths, only some
of which we do take. There are also many measures we could take, that
would raise the number of such deaths, some of which we do take. Fur-
thermore, it is not obvious that we are wrong to fail to do all we can to
reduce the number of deaths. For the purposes of this discussion, I will
focus on just one aspect of this failure, the failure to impose a national
speed limit of 50 miles per hour in the US.
If there were a national speed limit of 50 mph, it is overwhelmingly
likely that many lives would be saved each year, as compared with the
current situation. One of the costs of the failure to impose such a speed
limit is a significant number of deaths. The benefits of higher speed
limits are increased convenience for many. Despite this, it is far from
obvious that the failure to impose a 50 mph speed limit is wrong.24 In
fact, most people believe what I will call lives for convenience:
23. According to a study by the Natural Resources Defense Council, "Some 64,000
Americans may die prematurely each year because of air pollution" (Reuters: May 9, 1996).
24. There are those who react to what I say about the 50 mph speed limit by declaring
that I have convinced them that it is wrong not to impose it. But what I say about the 50
mph speed limit can also be said about a 40 mph speed limit, or a 30 mph speed limit,

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
Lives for convenience: We are not morally obligated to impose a na-

tional speed limit of 50 mph (or less).
If we reject life for headaches as obviously wrong, we must find a morally

significant difference between it and lives for convenience. I will devote
the remainder of this article to examining the most promising candi-
dates for such a difference.25
(i) The victims of higher speed limits are unknown, whereas life for
headaches envisages the death of an identifiable victim. Even if we
could tell how many more people die because of higher speed limits, we
could not tell which of the victims would not have died, had the limit
been lower. We cannot, therefore, point to specific individuals as the
ones whose lives were the price for our increased convenience.
It is hard to see how this epistemic difference could make a moral
difference. Consider the following two scenarios: A: New York City is
struck by a series of freak accidents involving power plants, which result
in the deaths of almost everyone within a twenty-mile radius of the Em-
pire State Building. The governor is informed that there are a handful of
survivors. He orders the national guard to round up the survivors, make
a record of their names, and kill them. B: In response to the same infor-
mation as in the previous example, the governor orders a hydrogen
bomb to be detonated in New York City, thereby eliminating the possi-
bility of discovering who survived the initial accidents. I hope it is clear
that at least part of what is morally wrong with the governor's actions
in each case is that some people are killed who would otherwise have
lived. It makes no difference that in case B it is impossible to tell which
people were killed by the nuclear explosion.
(ii) It is not just that we cannot tell whose deaths are attributable to
or even about abolishing private automobiles altogether. Very few are hardy enough to
follow their respect for life to such extremes.
25. The discussion that follows is similar, in some respects, to the more detailed discus-
sions in Peter Unger's recent book, Living High and Letting Die (Oxford: Oxford University
Press, 1996). We are both concerned to identify factors that might account for different
intuitive judgments on a pair of cases. Unger's cases involve allowing considerable harm
to befall others at the benefit of increased convenience, or even just avoiding minor ex-
pense, of one. Unlike my examples, Unger's do not address the possibility of aggregating
minor benefits to outweigh major harms. Unger is also more concerned than I am to
identify psychological factors that influence differential judgments. One of his aims is to
strip such factors of their force by exposing them. I only discuss factors that have at least
an outside shot at being morally relevant. Our overall aims are similar, though. We both
argue that the initial intuitive judgments about our cases are mistaken.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
the higher speed limits, it is that there is no fact of the matter as to who
would not have died, had the limits been lower. A counterfactual of the
form 'x would not have died had the speed limit been lower' has no
determinate truth value. Even if this is not so, it is, at least, indetermi-
nate who will die as a result of raising or not lowering the speed limit.
The weaker of these claims clearly has no moral relevance. Consider
these two prospective laws, to be considered by the Texas State Legisla-
ture: I: two victims are to be picked at random from the electoral roll one
day after the passage of this law. They are to be executed two days later.
D: Fred and Barney, whose names were picked at random from the elec-
toral roll a few minutes ago, are to be executed three days after the pas-
sage of this law. The fact that, at the time of passage, it is indeterminate
who will die as a result of I, but determinate who will die as a result of
D, obviously makes no moral difference.
Even if there is never any fact of the matter as to whose deaths are the
result of higher speed limits, this makes no moral difference. Consider
the following scenario: Two prisoners of conscience, Smith and Jones,
are slated for execution in a small totalitarian republic. The president,
Shrub, troubled by the effect of an Amnesty International campaign on
his public image, decides to spare one of the prisoners. He cannot make
up his mind whose life to spare, so he devises the following apparatus:
Smith and Jones are placed in separate cells, each with air vents leading
to a canister of poison gas, which is set to release its contents at noon.
A computer is programmed to select a three-digit number at random at
one second before noon. If the number is even or zero, the computer
will close the air vent in Smith's cell; if the number is odd, the computer
will close the air vent in Jones' cell. The random number selection proc-
ess is truly indeterministic. The vice president of the republic, Fowl,
does not approve of Shrub bowing to liberal pressure. Fowl unplugs the
computer at one minute before noon. The gas is released at noon, and
both Smith and Jones die. I hope it is clear that Fowl has done some-
thing bad in this example. Two people have died instead of one. It is also
clear that there is no fact of the matter as to which prisoner has been
killed by Fowl's actions. Since the number selection process is truly in-
deterministic, there is simply no fact of the matter as to whether the
number would have been even or odd. There is no nonarbitrarily identi-
fiable victim of Fowl's action, and yet it is morally on a par with killing
one person.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
Even though neither (i) nor (ii) can ground a moral difference be-
tween life for headaches and lives for convenience, they might explain a
psychological difference between them. We are more inclined to give to
charities when we are told of specific individuals who will benefit from
our gift. When the story of a child, whose parents cannot afford his life-
saving operation, is aired on television, complete with interviews with
the child and parents, donations flood in. When we are told of a natural
disaster, that threatens the lives of thousands of children, our donations
may even be less than in the previous case, and they will almost cer-
tainly not be thousands of times greater. Clearly, however, the life of the
one child is no more morally significant than the lives of any of the
others. If consequentialism's commitment to life for headaches is to be
a strike against the theory, we must do better than (i) or (ii).
(iii) Speed limits are set by the government, which is elected by the
people. So higher speed limits are the end product of a democratic proc-
ess, in which the victims participate. In a sense, then, the victims of
higher speed limits have chosen the system that kills them. This is not
true of the prospective victim of life for headaches.
First, many of the victims of higher speed limits do not participate in
the democratic process. In the 1994 US congressional elections, only 59
percent of registered voters participated, and only two-thirds of eligible
voters were registered. Thus, only about 39 percent of eligible voters
participated. But, it might be objected, the other 61 percent could have
participated, had they chosen. True, but many of the victims of higher
speed limits are not even eligible to vote, since they are children. Sec-
ond, the question of speed limits was not a campaign issue, so many
voters who did participate did not know their chosen candidate's stand
on the issue. Third, it is also possible that many voters had to choose
between candidates, all of whom supported higher speed limits, so they
did not even have the option to vote against them. Fourth, it is highly
probable that some of the victims of higher speed limits did vote for
candidates who opposed such limits.
(iv) Even though (at least some) victims of higher speed limits cannot
be said to have chosen the system that kills them, at least they have
benefited from the system, along with all the other automobile users.
Once again, this is most likely not true of all victims. Some may be
drivers or passengers in vehicles that do not take advantage of higher
speed limits. Some may be pedestrians. Some may never even approach

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
an automobile, and yet be killed by the higher levels of particulate pollu-

tion caused by the greater consumption of gasoline that results from
higher speed limits. Besides, our intuitive reaction to life for headaches
would not change if it were specified that the victim was one of the many
headache sufferers, and that her headache would be relieved before she
were killed.
(v) The victims of higher speed limits, even though they may have
neither chosen nor benefited from those limits, have at least freely cho-
sen to undertake the risk of being harmed. The dangers of driving, or
being driven, are well known. Those who choose to travel by road, there-
fore, are at least partly responsible for any harm that befalls them. The
same cannot be said of the prospective victim of life for headaches.
Once again, this is clearly not applicable to the victims of higher levels
of particulate pollution. Nor does it seem to be true of all the victims of
road accidents. Many children are killed on the roads. Many of these
may have had no say over whether they were to travel that way. Perhaps
we will say that their parents voluntarily assumed the risk on their be-
half. But this seems to be an inadequate reply. Would our reaction to life
for headaches be significantly different, if the prospective victim were a
child, chosen at random from among those whose parents had agreed
to the selection procedure? Besides, it is not clear how free is the choice
to travel by road, even for well-informed adults. For many of the victims
of road accidents, the alternatives may have been excessively burden-
some, if not nonexistent. Many people do not have access to basic serv-
ices, such as groceries and health-care providers, except by road.26
(vi) We do not know for certain that more people die on the roads (or
from particulate pollution) when speed limits are higher. The evidence,
though inductively strong, is not conclusive. Therefore, there is some
chance, albeit small, that a decision to raise (or not lower) speed limits
will not result in more deaths. Even if the alternatives being considered
were the total abolition of private cars versus the status quo, it is possible
that the former would not save any lives over the latter. Life for head-
aches, however, deals with the death of an innocent, not simply the
(overwhelming) likelihood of death.
Certainly it is possible that higher speed limits will not result in more
26. For an interesting discussion of this point, and others relevant to the current discus-
sion, see Guido Calabresi, Ideals, Beliefs, Attitudes, and the Law (Syracuse: Syracuse Uni-
versity Press, 1985), Chap. 1.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
deaths than lower ones. Perhaps drivers drive more carefully at high
speed, or car manufacturers work harder to develop safety features
when speed limits are higher. It is even possible that the abolition of
private cars would not save lives. The chances, however, are exceedingly
slim. Can the difference between certainty and near certainty of death
really mark a morally significant difference between life for headaches
and lives for convenience? I think not. The seeming counterintuitiveness
of life for headaches would hardly be diminished if we specified that the
victim were merely to be shot in the head at close range with a Magnum
.44. There would, in that case, be a small, but finite, probability that she
would not die. Or perhaps the victim could be made to play Russian
roulette with a thousand-barreled revolver with only one empty cham-
ber. Such alterations do not change most people's intuitive reactions to
life for headaches.
(vii) When we raise (or fail to lower) speed limits, we do not kill any-
one. The most we do to the victims of higher speed limits is to let them
die. Life for headaches envisages killing an innocent victim. Since there
is a morally significant difference between killing someone and letting
them die, there is a morally significant difference between life for head-
aches and lives for convenience.
Whether the distinction between killing and letting die is morally sig-
nificant is highly controversial. There is barely any agreement on what
the distinction amounts to, let alone whether it can bear moral weight.27
However, let us assume, for the sake of argument, that we understand
what the distinction is and that it is morally significant. Have we now
found the difference that allows us to say that life for headaches is obvi-
ously wrong, whereas lives for convenience is not wrong? That depends
on just what the moral significance of the killing/letting die distinction
is supposed to be. Perhaps it is worse to kill someone than to let them
die, but only proportionately worse. Perhaps, that is, it is as bad to kill
one person as to let, say, ten people die, or twenty, or a hundred. On the
other hand, perhaps it is lexically worse to kill than to let die. That is, it
is worse to kill even one person than to let any number of people die.
Only if killing is lexically worse than letting die might the killing/letting
27. For more on these questions, see Bonnie Steinbock and Alastair Norcross, eds., Kill-
ing and Letting Die (New York: Fordham University Press, 1994), and Jonathan Bennett,
The Act Itself (Oxford: Oxford University Press, 1995).

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
die distinction do the work we need it to do. Suppose it is only propor-

tionately worse to kill than to let die. If we can justify letting someone
die by the resulting gains in convenience to many, we can justify killing
someone by the resulting gains to proportionately more. If the signifi-
cant difference between lives for convenience and life for headaches is
that the latter involves killing, but the former involves letting die, and
lives for convenience is correct, then we can justify letting some people
die by the resulting gains in convenience to many. So, just how defensi-
ble is the claim that killing is lexically worse than letting die? If it is true,
it follows that it is worse to kill one person than to let the entire sentient
population of the universe (including the one person) die. This is clearly
unacceptable.
(viii) When we raise (or fail to lower) speed limits, we do not intend
any of the deaths that we thereby bring about, even though we may
foresee them. Life for headaches envisages the intentional killing of an
innocent victim. Since there is a morally significant difference between
bringing about a death that is intended and bringing about a death that
is merely foreseen,28 there is a morally significant difference between life
for headaches and lives for convenience.
Once again, the questions of what the intending/foreseeing distinc-
tion is and whether it can bear moral weight are controversial.29 Let us
assume, again, that we understand the distinction, and that it is morally
significant. It is clear, for the reasons given above with respect to the
killing/letting die distinction, that bringing about a death that is in-
tended will have to be lexically worse than bringing about one that is
merely foreseen, if the distinction is to mark the right sort of difference
between lives for convenience and life for headaches. But, once again,
this is an untenable position. Could it really be worse to bring about and
intend the death of one person than to bring about, but merely foresee,
the deaths of everyone, including the one?
There is another possibility to consider. Even though killing may not
28. The moral significance of this distinction is often defended in the context of the
Doctrine of Double Effect, according to which the distinction between evil that is an in-
tended means to a good end and evil that is a foreseen side effect of achieving a good end
is crucial.
29. For more on these questions, see, among others, Steinbock and Norcross, Killing
and Letting Die, esp. Chaps. 6-9, 20, and Introduction to the 2nd ed., and Bennett, TheAct
Itself.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
be lexically worse than letting die, and intending a death may not be
lexically worse than foreseeing it, perhaps intentionally killing may be
lexically worse than merely foreseeing the death of someone you only
let die. Since, it may be argued, we neither intend the deaths of the
victims of higher speed limits, nor do we kill them, we must consider the
two distinctions working together. This suggestion, however, is just as
implausible as the previous ones. Consider the following scenario:
Jane, Dick, and the fate of humanity: Jane is a crack shot with a high-
powered rifle, who has just discovered a nefarious plot against all hu-
manity. A bomb, powerful enough to destroy all life on Earth, is con-
nected to an ingenious trigger device. The device, hidden somewhere
in Disneyland, is set to detect any signs of human life within the
amusement park. If it detects any signs at noon, it will detonate the
bomb. It had been thought that the park was clear, but, at the last
minute, Dick, who is absent-minded and hard of hearing, is discov-
ered several hundred yards inside the perimeter. It is now ten seconds
before noon, and the onLy way to prevent the bomb exploding is for
Jane to shoot and kill Dick with her high-powered rifle. If she shoots
him, she will intentionally kill him. If she does not shoot, she will
foresee the deaths of all humanity, whom she will let die.
Clearly, it cannot be worse for Jane to shoot Dick than to allow all living
things on Earth, including Dick and herself, to be killed by the bomb, even
though she would be neither killing them nor intending their deaths.
There is another reason why even a combination of the killing/letting
die distinction and the intending/foreseeing distinction cannot ground
the right sort of difference between lives for convenience and life for
headaches. A version of life for headaches, in which the victim is let die
and her death is merely foreseen, would not seem significantly less
counterintuitive, at least from the point of view of common-sense mo-
rality. Consider this scenario:
Drowning baby: Fred hears of an outbreak of minor headaches, last-

ing from 2 until 3 every afternoon, in a neighboring town. He happens
to possess the only known cure, which, if he gets it to the town before
2, will prevent all the headaches. Otherwise the people will suffer for
an hour. Fred is hurrying to catch the only train that will get him there
in time, when he passes a baby drowning in a pond. If he stops to

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC
rescue the baby, which he can accomplish easily, he will miss the
train, and all the people in the neighboring town will suffer a minor
headache for one hour. If he does not stop, the baby will die, but Fred
will neither have killed her nor intended her death.
Now consider the following claim:
Life for headaches 2: there is some finite number of headaches, such

that it is permissible for Fred to abandon the baby to a watery grave
in order to avoid them.
I think it is clear that whatever counterintuitiveness attaches to life for

headaches also attaches to life for headaches 2. At the very least, it is hard
to see how one could judge the former to be obviously wrong and the
latter to be right.
Despite the seeming counterintuitiveness of life for headaches and the
inoffensive nature of lives for convenience, I have been unable to find a
difference between them that would justify the claim that one is false
and the other true. There may yet be such a difference, but I think I have
looked hard enough, for someone who does not believe life for head-
aches to be false. I now pass the burden of discovering such a difference
to those who stubbornly persist in believing that life for headaches is not
only counterintuitive, but false. In the meantime I conclude that conse-
quentialisims commitment to life for headaches is not a decisive strike
against the theory.

130.63.180.147 on Thu, 10 Sep 2020 21:10:35 UTC

week04_Norcross1997

Uploaded by

Copyright:

Available Formats

week04_Norcross1997

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

week04_Norcross1997

Uploaded by

Copyright:

Available Formats

Comparing Harms: Headaches and Human Lives

Author(s): Alastair Norcross

Stable URL: http://www.jstor.com/stable/2961948

This content downloaded from

Consequentialists are sometimes unsettled by the following kind of ex-

This content downloaded from

In order to avoid life for headaches, therefore, a consequentialist must

Incomparable: The state of affairs constituted by any number of fairly

To say that a state of affairs (or option) A is incomparable with a state

1. Of course, it is one thing to claim that such a deontological restriction exists, it is

This content downloaded from

of headaches is most likely to do so because she thinks it worse to kill

Con 1: An act is permissible iff it is at least as good as all the available

Con 2: An act is permissible iff there are no better alternatives.

If we accept Incomparable, and consider the situation in which we ei-

A more promising tactic for a consequentialist who wishes to avoid life

Less: The state of affairs constituted by any number of fairly minor

A consequentialist who accepts Less denies that it is ever permissible to

This content downloaded from

Less' is wildly implausible. Roughly, it says that anything worse than a

This content downloaded from

Even if we accept S, we may yet attempt to hang on to Less. Recall that

This content downloaded from

Intransitive Trolley: A three-car trolley is speeding out of control to-

This content downloaded from

Is Less such a compelling principle that we should reject something as

6. Philosophical Studies 59 (1990): 79-90.

This content downloaded from

quentialist, such as a utilitarian, with a reason to reject the transitivity

This content downloaded from

stitute an objective difference. If so, it is possible that his comfort level

We are ignoring the effects of "triangulation": The self-torturer can

His response is to stipulate, again, that he is interested in a case in which

But, empirically speaking, his preferences as between s and o can ex-

Imagine the self-torturer comparing successive settings with o. In front

This content downloaded from

This content downloaded from

Claim 1: for any unpleasant or "negative" experience, no matter what

Claim 3: a mild discomfort for the duration of one's life would be

Claims 1-3 all "depend on an appropriate 'other things equal' clause"

This content downloaded from

This argument is provocative, and initially c

This content downloaded from

nail throughout? But in comparing A with Y we are supposed to imagine

Consider the following spectrum: red, reddish orange, orange, orang-

If we could get from excruciating torture to mildly annoying hangnail in

This content downloaded from

not ready to trust my (or Temkin's) intuitions on the comparison of A

This content downloaded from

This content downloaded from

This content downloaded from

This content downloaded from

16. Temkin seems to endorse this explanation on p. 194.

This content downloaded from

[I]n essence, I think significant amounts of torture have lexical prior-

This content downloaded from

This content downloaded from

This content downloaded from

ling, can coexist happily with claim 4, and w

Claim 3: a mild discomfort for the duration of one's life would be