Scanlon on Double Effect
Ralph Wedgwood
Merton
College, University of Oxford
In this new book
Moral Dimensions, T. M. Scanlon
(2008) explores the ethical significance of the intentions and motives with
which people act. According to Scanlon, these intentions and motives do not
have any direct bearing on the permissibility of the act. Thus, Scanlon
claims that the traditional Doctrine of Double Effect (DDE) is mistaken.
However, the way in which someone is motivated to act has a direct bearing on
what Scanlon calls the act’s “meaning”. One particularly important kind of
“meaning” that an act can have consists in the ways in which it is appropriate
for various people to blame the agent
for the act. So the book ends with an extended analysis of blame and
blameworthiness.
As anyone
acquainted with Scanlon’s work would expect, the book is full of extremely valuable
insights. The discussion of blame in the last chapter (122–214) is of
especially great and lasting importance. In these comments, however, I shall
follow the common practice of concentrating on the areas on which I disagree
with the work that I am discussing. In particular, I shall defend the DDE
against Scanlon’s attack. This attack has three parts: first, he argues
directly against the plausibility of the DDE (18–20 and 56–62); secondly, he
offers an explanation of why moral thinkers might have been misled into
accepting the DDE (20–28); and thirdly, he offers an alternative account of the
main intuitions that philosophers have adduced in favour of the DDE – by
arguing that we can do justice to these intuitions without appealing to the DDE
(28–36).
I shall not
have time to answer the third part of Scanlon’s attack here.[1] Instead,
I shall focus exclusively on the first two parts of his attack. First, however,
I shall comment briefly on the question of how the DDE should be interpreted.
1. The interpretation of the DDE
Scanlon
gives the following formulation of the DDE (1):
The doctrine of double effect holds that an action that aims
at the death of an innocent person, either as its end or as a means to its end,
is always wrong.
The version
of the DDE that I shall defend here is slightly weaker than this formulation. First,
my version of the DDE is restricted to acts that actually do have a bad event (like the death of an innocent person)
as one of their effects. In my view, acts that aim at a bad event but fail to
achieve their aim fall into an importantly different category (although
unfortunately I shall not be able to discuss such unsuccessful acts here).
Secondly,
the version of the DDE that Scanlon discusses is absolutist: it implies that acts that have an innocent person’s
death as one of their intended effects are always
wrong. But Warren Quinn’s (1989) version of the doctrine seems more plausible:
according to Quinn’s version, the fact that an act has an innocent person’s death
as one of its intended effects does not invariably make the act
impermissible, since sufficiently strong countervailing reasons may make the
act permissible after all. Instead, this version of the DDE implies only that,
other things equal, there is a stronger reason against an act if a
bad event (like an innocent person’s death) is one of the act’s intended effects than if it is merely
an unintended effect.
Whenever we
judge that it is “impermissible” for an agent to act in a certain way, or that
there is “a reason” against the agent’s acting in that way, we seem to be
making a judgment about an act-type,
in relation to the agent’s situation at the relevant time. In theorizing about
act-types, there is a complication that we need to take account of: some
act-types are more specific than others.
One way in
which one act-type can be more specific than another is by incorporating a
specific intention. Thus, one relatively general act-type is diverting the
runaway trolley onto the sidetrack. Relatively specific act-types include diverting
the trolley with the intention of saving the five people on the main track, and diverting the trolley with the
intention of killing the person on the sidetrack. I shall label
act-types that do not incorporate the intention in this way “thin” act-types,
while act-types that do incorporate these intentions will be called “thick”
act-types.
I shall assume
here that the DDE is primarily concerned with thick act-types. (Unfortunately,
I shall not have time to explain how to apply the doctrine to thin act-types.)
The central claim of the DDE, as I interpret it, is that, other things equal, there
are stronger reasons against a thick act-type that involves successfully
executing an intention to bring about a bad event (like an innocent person’s
death) than against any otherwise similar thick act-type that does not involve such
an intention.
What is
meant here by “intention”? Strikingly, Scanlon does not distinguish between the
intention with which an action is
done and the motivation of the action
(10f.). The motivation of an action consists of the whole mental process that results
in the action: this process typically includes not only intentions but many
other motivating states as well – such as desires, emotions, wishes, and
beliefs. Intentions directly control
and guide behaviour, whereas the other motivating states control behaviour only
indirectly – by means of leading one to have certain intentions. The
action itself is the execution of the
intention; the agent’s intention is an essential constituent of the action
itself (whereas the other motivating states only explain the agent’s having this intention).
As I
understand it, the DDE implies that intentions have intrinsic or non-derivative
ethical significance. There is a stronger reason against executing an intention
to bring about a bad event (like an innocent person’s death) than against
bringing about such an event without intending it precisely because of
the difference that consists in the presence or absence of this intention. The
greater strength of the reason against acting is explained, not by something
that is merely correlated with the intention, but by the intention itself.
2. Scanlon’s objections to the DDE
Scanlon’s first objection consists simply of the following terse remark: “To my knowledge no one has come up with a satisfying theoretical explanation of why intention, in the sense that is involved here – the difference between the consequences that are intended and those that are merely foreseen – should make a moral difference” (18). But several philosophers have attempted to give such a theoretical explanation of the DDE. For example, according to Thomas Nagel (1986, 181), the explanation of why there is an especially strong reason against your executing an intention to bring about a bad event (like an innocent person’s death) is that in such cases, your “will” is being “guided by evil”. I believe that Nagel’s account can be developed to give a plausible explanation of the DDE (see Wedgwood 2009, 334–6). So in my view, we should not be persuaded by Scanlon’s first objection to the DDE.
Scanlon’s second objection is based on Thomson’s (1985) “Loop” case. As Scanlon puts it (18):
There are cases in which applying the [DDE] seems to give the wrong answer. For example in the well-known trolley-problem case it seems permissible to switch a runaway trolley onto a sidetrack on which it will hit only one person rather than allow it to continue straight ahead and hit five. But it also seems permissible to switch the trolley in the “Loop” case, proposed by Judith Thomson, in which the sidetrack loops around and rejoins the main line, so that if the trolley does not hit the one person and thereby come to a stop, it will continue around the loop and hit the five from the other side.
It seems to me, however, that there is a stronger reason against diverting the trolley in the Loop case than in the original Trolley case. In the Loop case, you need the trolley to hit the person on the sidetrack for your plans to succeed: if the person on the sidetrack gets out of the path of the trolley before it can collide with him, your plans will have failed. In the original Trolley case, by contrast, it is entirely inessential to your plans that the trolley will collide with the person on the sidetrack: if the person on the sidetrack escapes from the path of the trolley before it can hit him, your plans will not have failed. So the two cases seem significantly different.[2] (This is not to say that diverting the trolley in the Loop case is positively impermissible, but only that there is a stronger reason against diverting the trolley in the Loop case than in the original Trolley case, which either makes diverting the trolley impermissible, or at least takes it significantly closer to the line that divides permissibility and impermissibility.)
Finally,
Scanlon endorses another one of Thomson’s objections to the DDE (19–20):
Suppose you were prime minister, and the commander of the air
force described to you a planned air raid that would be expected to destroy a
munitions plant and also kill a certain number of civilians, thereby probably
undermining [the enemy’s] public support for the war. If he asked whether you
thought this was morally permissible, you would not say, “Well, that depends on
what your intentions would be in carrying it out. Would you be intending to
kill the civilians, or would their deaths be merely an unintended but
foreseeable (albeit beneficial) side effect of the destruction of the plant?”
Scanlon does
not explicitly justify his claim that if you were prime minister, you “would
not say” this. But his reason for this claim seems to be that he sympathizes
with Thomson’s (1991, 293) rhetorical question: “Can anyone really think that
the [would-be bomber] should decide whether he may drop the bombs by looking
inward for the intention with which he would be dropping them if he dropped
them?”
Thomson
assumes here that applying the DDE in deciding what to do in this case would
involve “looking inward”, to discover what intention one would be acting with
if one dropped the bombs, and basing one’s decision about whether or not to drop
the bombs on this purely psychological information about one’s own state of
mind.
However,
this assumption is a travesty of the DDE. What the DDE says about this case is
that there are stronger reasons against the thick act-type dropping the bombs with the intention of killing civilians than
against the thick act-type dropping the
bombs with the intention of destroying the munitions plant. So the DDE supports
the conclusion that the former act-type is impermissible, while the latter is
permissible. But this conclusion does not
imply that what the bomber should do is determined by any purely psychological
facts about the intentions that he would
have if he dropped the bombs. It simply implies that he ought not to drop the
bombs with the intention of killing the civilians. It is irrelevant if he has
the property that if he were to drop the bombs, he would do it with the
intention of killing the civilians – after all, that is presumably a
property that he ought not to have.
So you
should respond to the bomber as follows: “You may drop the bombs, but only with
the aim of destroying the munitions plant. You may not drop the bombs with
the aim of killing the civilians; that would be wrong.” If the bomber retorts,
“Actually, I hate the enemy so much that were I to drop the bombs, I would be
doing it with the intention of killing their civilians”, then you should reply:
“For the purposes of answering your question, I am not interested in your
despicable attitudes towards enemy civilians. I am telling you what plan you
may adopt. You may adopt the plan of
destroying the munitions plant by means of dropping the bombs, and you may not adopt the plan of killing the
civilians by means of dropping those bombs: the plan that you may adopt would
still be carried out if your dropping the bombs somehow destroyed the munitions
plant without killing the civilians; it would not be carried out if your dropping
the bombs killed the civilians without destroying the munitions plant.”
In general, one
can make choices that ensure compliance with the DDE without ever “looking
inwards” or thinking about one’s own intentions at all. One need only think,
not about simple options (like dropping
the bombs), but about options that have a more complex structure (like destroying
the munitions plant by means of dropping the bombs).
Suppose that
the bomber chooses the option of destroying the munitions plant by means of
dropping the bombs. Since a choice is
the formation of an intention, the bomber has thereby formed the intention to
destroy the munitions plant by dropping the bombs (not the intention to kill the civilians by dropping the bombs). So
if he executes this choice, he will be complying with the DDE, by dropping the
bombs with the appropriate intention. Deliberating about these more complex
options thus allows us make choices that ensure compliance with the DDE, even without
“looking inwards” or thinking about our intentions at all.
We can
presumably add sufficient details to this case so that the would-be bomber
really ought to drop the bombs with
the intention of destroying the plant. The DDE then implies that the only permissible course of action is to
drop the bombs with the intention of destroying the plant, while it is impermissible to drop the bombs with the
intention of killing the civilians.
Scanlon objects
that this implication of my version of the DDE is incompatible with the
fundamental connection between permissibility and choice. As he says (58f.), “the question of
permissibility … applies only to alternatives between which a competent
agent can choose”:
The suggestion is that it might be impermissible either to
bring about a result with certain bad reasons in mind or to fail to bring it
about at all, and that the only thing that would be permissible would be to
bring it about for the right reasons. If I am correct about the connection
between permissibility and choice, this makes sense only if acting for those
different reasons is something that the agent can choose to do. I do not
believe that such a choice is possible.
Scanlon’s formulation
of this objection reflects his failure to distinguish between the intentions
with which a person acts and the person’s motivating
reasons for her action.
However, this point does not matter here, since it is equally plausible that the
agent also cannot form an intention by directly choosing to form that intention.
Scanlon’s claim
that there is a fundamental connection between permissibility and choice also
seems plausible. However, this connection need not take the precise form that he
describes. There is a slightly weaker connection worth considering: viz., if a
course of action is the only permissible option available, there must be some
choice that the agent can make such that, necessarily, if the agent executes
that choice, she will perform that permissible course of action. The difference
is between the permissible action’s being the object of a possible
choice, and its being the necessary consequence of the execution of a
possible choice.
If only this
slightly weaker connection holds between permissibility and choice, there is no
problem for the DDE. Even if you cannot choose your intentions, you can
choose to destroy the munitions plant by dropping the bombs (as opposed to
choosing to kill the civilians by dropping the bombs). If you execute this choice,
you will be complying with the DDE – dropping the bombs with the appropriate
intention.
Given the
intuitive plausibility of the DDE, we should prefer this slightly weaker
connection between permissibility and choice, on the grounds that it creates no
problems for the DDE. We can thereby do justice to all the intuitions that support
Scanlon’s claims while rejecting his criticisms of the DDE.
3. Scanlon’s explanation of the “illusory appeal of double effect”
Scanlon
suggests that the illusory appeal of the DDE derives from a confusion between
what he calls the “critical” and the “deliberative” employment of moral
principles (20–28). A moral principle is employed deliberatively when it is
used to evaluate which of the acts between which the relevant agent can choose
are permissible and which are impermissible. A principle is employed critically
when it is used to judge how well a certain agent has reasoned in making his
choices (that is, in forming his intentions) about what to do – e.g. in assessing
whether the agent is blameworthy.
Scanlon’s
conception of “employing a moral principle” is of considerable importance for
understanding his positive moral theory, but the details of his conception will
not matter for our present purposes. The central point is just that although an
agent’s intentions are normally relevant to determining how well the agent has
reasoned in deciding what to do, and whether the agent counts as blameworthy,
these intentions are not relevant to the permissibility of the act. So
Scanlon’s point is at least closely akin to the relatively familiar idea that
advocates of the DDE confuse the blameworthiness of the agent for having
a certain intention with the impermissibility of the act.
We should
certainly agree that these are not equivalent. An act can be impermissible even
if the agent is not blameworthy in any way, for the simple reason that the
agent may have an excuse. (E.g. the agent may have been blamelessly
ignorant of factors that made the act impermissible.) But what Scanlon needs is
not the thesis that impermissibility does not entail blameworthiness, but the
converse – the thesis that blameworthiness does not entail
impermissibility.
There do
seem to be cases in which an agent is blameworthy for some aspect of the way in which she was motivated to act, but the
act itself is entirely permissible. In these cases, however, what is the agent blameworthy
for? It surely cannot be for simply performing
an act of this act-type (since this act-type is entirely permissible), or for
the intention with which she acted (since the intention is an essential
constituent of this entirely permissible act-type). So if the agent’s act is entirely
permissible, she must be acting with an entirely innocent intention. If she is
still blameworthy, it can only be for some other
aspect of the motivational process – such as her desires or emotions or
beliefs, and the way in which these motivating states that led her to have that
intention – not for the intention itself.
So it seems
hard to imagine how anyone could be genuinely blameworthy for having a certain intention unless there is something
intrinsically wrong or impermissible about that intention. But an act just is
the execution of an intention; the intention is an essential constituent of the
act (while the motives that led the agent to have that intention are not
essential constituents of the act in the same way). So if an act is the
execution of an intrinsically impermissible intention, the defect in the
intention must surely be inherited by the act itself. So there do not seem to
be any clear cases in which (i) the agent is genuinely blameworthy, not merely
for the motives that led her to have the intention with which she acts, but for
the intention itself, and yet (ii) acting with that intention is not wrong or
impermissible in any way.
In short, even
if Scanlon is right that some advocates of the DDE confuse the blameworthiness
of an intention with the impermissibility of acting on the intention, this does
not in any way undermine the DDE. If one is blameworthy for the intention with
which one acts, then the act itself will also be impermissible.
References
McMahan, Jeff (2009). “Intention,
Permissibility, Terrorism, and War”, Philosophical Perspectives 23: 345–72
Nagel,
Thomas (1986). The View from Nowhere
(Oxford: Clarendon Press).
Quinn, Warren (1989). “Actions,
Intentions, and Consequences: The Doctrine of Double Effect”, Philosophy and
Public Affairs 18: 334–351.
Scanlon, T. M. (2008). Moral Dimensions: Permissibility, Meaning,
Blame (Cambridge, MA: Harvard University Press).
Thomson, Judith Jarvis (1985). “The
Trolley Problem”, Yale Law Journal 94: 1395-1415.
———— (1991). “Self-Defense”, Philosophy
and Public Affairs 20: 283–310.
Wedgwood, Ralph (2009). “Intrinsic
Values and Reasons for Action”, Philosophical
Issues 19 (2009): 342–63.
[1] For an effective rebuttal of this part of Scanlon’s attack, see McMahan (2009).
[2] According to Thomson (1985, 1403), “we cannot really suppose that the presence or absence of that extra bit of track makes a major moral difference.” But the presence of this “bit of track” obviously makes a major difference to which counterfactuals are true in this case; and since in this case you know which counterfactuals are true, it also makes a major difference to the intentions with which you act.