The Philosophical Quarterly Vol, 42 No. 168
ISSN 0031 ~ 8094 32
A SOLUTION TO THE PROBLEM OF MORAL LUCK
By Bryxaor Browne
Praising, blaming, rewarding and punishing, are, in practice, justified
by reference to what the person who is treated in these ways has done;
but we soon become worried about the fairness of treating people on the
basis of what they do when we see that chance has played a crucial role.
How should we treat those whose actions are only different because of
the difference made by chance? If, for example, we treat the murderer
in the same way as we treat the man who attempts murder then weseem
tolack an appropriate response to their different actions; but if we treat
them differently, are we justified in doing so because of a difference
made by chance?
The condition of control,' that is, the idea that a person is only
responsible for that which is under his control, is central to the account
of responsibility which gives rise to the problem of moral luck. It has
seemed both impossible to deny (because it is counter-intuitive to hold
a person responsible for what is independent of his control) and
impossible to sustain (for reasons which I shalll give). However, I shall
argue that the condition of control is central only toa particular view of
moral agency which seems attractive only because it helps to justify
attitudes which we believe are central to morality but which are, in
fact, morally wrong. My aim will be to reconcile our conflicting
intuitions by accommodating them within a range of reactions to
wrong-doing which does not leave us with the problem of moral luck
But first I shall argue that the ‘condition of control’ isa misnomer and
as such it helps us to misconstrue the problem of moral luck.
It might be thought that the problem of moral luck arises from
situations where chance events can be said to be in some way in conflict
with an agent’s control, but care is needed if we are to discover the real
nature of the problem, A chance event may be said to be in conflict with
an agent’s control when it takes away control, as it might, for example,
in the case of a car which skids on a patch of ice. A driver who recklessly
' ‘This term is taken from Nagel’s paper ‘Moral Luck’, in Thomas Nagel, Mortal
Questions (Cambridge: Cambridge University Press, 1979346 BRYNMOR BROWNE
drives too fast may be unlucky in skidding on a patch of ice with the
consequence that he loses control of the car; if he kills someone as a
result, then he is, to some extent, morally unlucky. In this case chance
and control conflict in the most obvious way; but deprivation of control
is not necessary before an agent can be said to be morally unlucky or
lucky. If the patch of ice had melted prior to the arrival of the driver,
this too would be a matter of luck but there would be no interference
with the driver's control. Here we have two different ways in which
chance may be involved in a person’s action. In one case a chance
occurrence interferes with control; in the other it does not.
Consider a further example; that of a doctor who accidentally injects
a patient with a poisonous substance, believing it to be a medicine.
Someone else has put the wrong substance into the syringe. Normally in
such a case the doctor would not be blamed for his action. The
difference that chance makes to his action is not in itself something
which leads to the doctor being blamed for the outcome of his action.
Had the doctor been carrying out the operation unnecessarily for
reasons of personal profit, he would then be morally blameworthy for
what he did.
Itis important to keep in view the fact that the agent’s control need
not suffer actual interference before he becomes morally lucky or
unlucky. The doctor's control when he injects the patient is the same as
it would have been if the correct drug had been put in the syringe. Here,
chance has madea difference to the doctor’s action without affecting his,
control either positively or negatively, although we might say that
there remained some sense in which he had lost some control over his
action in that he was not doing what he thought he was doing or doing
the action he intended (injecting the patient with medicine). However,
if, by some farther fluke, someone drops the poisoned syringe after
which the correct drug is given to the doctor who then injects the
patient as intended, the doctor would then be lucky, in control, doing
what he thought he was doing and doing the action he intended.
We can see from these examples that agents can be said to be morally
lucky or unlucky if their actions have a kind of contingency. In these
cases chance may or may not interfere with an agent's control of his
action. Agents seem not to be responsible for these aspects of their lives
and actions and yet what they are held responsible for will be
determined in part by these differences
Some might try to insist that there is only a problem of moral luck in
those cases where chance takes away control and that we only imagine
that there is a problem in the other cases which involve contingency
However, the suggestion that we havea problem of moral luck in actionASOLUTION TO THE PROBLEM OF MORAL LUCK 347
cases simply because chance can or does interfere with control within
the action, is ill-founded. In normal cases of constitutive luck ~ for
example, luck in being the kind of person one is ~ there can be no
suggestion that the problem of moral luck arises through luck being in
conflict with an agent’s control over his action. The problem of moral
luck arises in such cases where the agent is in complete control and is
held responsible for his action, Similarly, an evil person who is powerful
may perform more evil actions than one who is weak and unable to do
as much. They may be equally evil in onesense, but can we nevertheless
treat them the same? Furthermore, even in those cases where we see a
problem in treating the attempt at murder less seriously than murder, a
problem arises about how we should treat both agents and not just the
‘one whose attempt is thwarted by chance; reflecting on the role played
by chance in the case of the attempted murder will lead us to think
again about the murderer.
‘One way in which people react (initially) to the problem of moral
luck is to try to tie responsibility to willings. Abandoning our intuitive
desire to react to what an agent does in favour of holding an agent
responsible for what he wills does not, of course, escape the problem of
moral luck, because an agent’s willings are not free of chance. What an
agent wills arises, in part, out of his situation and his character, over
which he must, at the very least, have less than complete control. Nor
can one separate the agent who wills from his character; his will must be
independent of his character to be luck-free but must be intimately
connected to a person ifit is to be his will. So we must either allow that
willings are not luck-free or we must completely separate the one who
wills from (his?) character.
T have suggested that our actions have a certain kind of contingency
which appears to threaten our responsibility for them; the presence of
this contingency seems to be incompatible with the kind of control
which we believe agents have when they are morally responsible. We
have seen that contingency is compatible with control, in its usual
sense. However, there is some way in which the presence of contingency
may still seem to conflict with control. Ifit does, then it looks as if the
kind of ‘control’ which we think is required must be one which is free
from contingency. Such a condition is not, properly speaking, a
condition of control, but more correctly a condition that the agent be
contingeney-free.
The attempt to find luck-free agency might be characterized as a
search for an uncaused cause of our actions as well as an agent whose
actions are not subject to causal influence after their execution. In our
attempt to adhere to the condition of control we slip into search for the348 BRYNMOR BROWNE
impossible. We have seen that what we do is always, in part, due to
factors which are not entirely ofour own making. Initially, the presence
of chance in our actions appears to threaten control. But care must be
taken here because ‘what is under our control’ is not the same as
‘entirely of our own making’. The first thought allows the presence of
chance in actions, in that the situation in which the agent has control is
made up of elements not of his own making. The man who drives a car
has it under his control, even though chance may have played a part in
his being hired to drive it. The latter thought represents a condition
which is unrealizable because whatever an agent does, he cannot act in
aworld entirely of his own making, or be an agent who is entirely of his
own making and who acts from options or possibilities entirely of his
own making. I will say that such a being would be a ‘complete
instigator’ of his actions.
Itis clear, then, that the ‘condition of control’ which gives rise to the
problem of moral luck is incoherent as an account of human agency and
that this requirement is really the requirement that an agent be a
complete instigator of his actions. However, the mere realization of thi
fact will not make the problem of moral luck go away. We have a deep-
rooted intuition which seems to tell us that the condition of control
(which I have analysed as the condition that an agent be a complete
instigator) is a requirement for moral agency. As Nagel observes in
“Moral Luck’, we do not simply notice that people who are responsible
are generally in situations where they have control and conclude that
control isimportant to responsibility, but we believe that the ‘condition
of control’ must be the basis on which we hold people responsible for
their actions. If this observation is correct (and | believe that it is), we
cannot explain our wish that agents’ actions be free of contingency in
terms of a slip which involves seeing contingency in our actions on the
model of interference with an agent’s control in the ordinary way (ice
on the road). Any solution to the problem of moral luck will need,
therefore, to explain why we should have such a deep-rooted, and yet
flawed intuition. It will also have to remove any tendency toseek such a
condition
Although the condition that an agent be a complete instigator of his
action meets with certain deep intuitions, we should not forget that we
are also strongly inclined to react to what agents do. We are even
inclined to regard ourselves as responsible, in certain ways, for actions
over which we have no real control. Ifa train driver accidentally, and
totally without fault, runs over a person lying on the tracks, he will feel —
and will be ~ responsible in some way for what he has done. Although
we might try to make him see that he was not at fault, we would,A SOLUTION TO THE PROBLEM OF MORAL LUCK 349
nevertheless, think that complete indifference on his part, on the
grounds that he was not to blame, would show a lack of sensitivity with
regard to the role he played in killing the person on the track. Again,
this example counts against the condition of control in that we think the
driver is responsible in some way which has moral relevance.
Of course, it will be said that the train driver may be responsible, in
some way, for what he did, but we would not hold him responsible in
the way that we would if he had deliberately killed; he is not
blameworthy. But if we return to the case of murder it is clear that we
think that we should react to a person who kills differently from the way
we should react to one who merely attempts to kill. Have we not, then,
abandoned what might be properly called ‘the condition of control’?
Surely, if the condition of control were so obviously correct, there
would be no temptation to treat people differently on the basis of the
outcomes of their actions. Even after we go through the arguments
showing that chance was responsible for the attempt becoming a
successful act, we resist the temptation to regard the action and the
attempt as morally equal. At the very least, we can say that we have
here competing intuitions; even if the condition of control could be
made coherent we would feel an inclination to say that we could not
ignore the difference (for example) between killing and not killing with
respect to what we expect from the relevant agent.
Let us now consider the position:
1. We want to react differently to people according to what they do,
including what they do as a result of the differences that chance
makes. But it seems to be unfair to hold a person responsible for
what he does as a result of differences that chance has made.
2. If we abandon the condition of control we seem to abandon an
intuitively plausible condition. But the condition of control is
incoherent — agents are not complete instigators.
We, therefore, need a way ofaccommodating our need to react to what
an agent does, including that which is contingent in his action or
beyond his control. But we also need, at the same time, some means of
removing the worry that seems to occur once we make this move. We
need something that reassures us that we do not treat people unfairly
when we hold them responsible for those actions in which chance plays
a crucial role. I suggest that all of this is possible, provided that we are
prepared to abandon certain attitudes to wrong-doing and certain
related practices
I will now argue that, by adopting different attitudes to wrong-doing
and alternative practices to punishment, the problem of moral luck is350 BRYNMOR BROWNE
overcome. I suggest that we shall be able to abandon the condition of
control, once we see that not only is it incoherent, but also that it only
suggests itself whilst certain attitudes are in play, attitudes which are, in
themselves, morally wrong. We soon go astray when we try to justify
them.
The actions of human agents require a wide range of appropriate
responses. A person who wills evil in his action must be treated
differently from one who does not, and this difference should be
reflected in our attitude when we blame him. We must react differently
to one who (for example) kills deliberately from the way we would to
one, such as the train driver, who kills accidentally or even carelessly
My position is that, whilst our attitude to evil actions must not be
hostile or against the agent, it must differ from our attitude in cases in
which there is no real evil intent on the part of the agent. I will argue
that this does not exclude the possibility of showing a wide range of
emotions which are not incompatible with a loving attitude
Many would say that anger is an appropriate reaction to wrong-
doing. Often, when people act out of anger they do things which they
later see as being wrong. Their anger is directed at someone and this is
often accompanied by a wish to do him some harm or to punish.
However, anger need not take quite this form. The kind of anger that I
have in mind is of the kind that a loving parent might feel on finding out
that his son had done something morally wrong. What would be the
right attitude of a parent to his son’s killing someone? Certainly, he
would blame him, but his attitude of anger, sorrow, concern, anguish
and perhaps sympathy would be compatible with loving his son. In this
way his attitude would not be against his son but for him. Let us suppose
that this kind of anger and sorrow is felt more generally, perhaps in the
context of a small close community; would, then, the problem of moral
luck arise in the differential treatment ofa murderer as against one who
attempted murder, where the relevant difference between their actions
is one of chance?
If we show more anger towards the man guilty of murder than
towards the man who attempts murder, can the murderer say: We are
as bad as each other in that we attempted the same action, so why are
you more angry at me?’ Can we justify the different treatment? It seems
to me that we would have no difficulty in justifying our different
treatment. Certainly, we must treat them differently because the
murderer must (albeit through chance) live with more anger. Also, we
must treat him differently because he has killed a man and this requires
a response. We do not need to address the question of fairness from the
old perspective: we do not feel hostility towards him and we no longerA SOLUTION TO THE PROBLEM OF MORAL LUCK 351
want to punish him, Even if his situation is mainly the result of bad luck,
it still demands a response and that response must recognize his
situation. Soitis reasonable to react differently, provided the reaction is
ofa certain kind.
‘My suggestion is, then, that the complaint which is made about the
unfairness of our reactions to actions where chance plays a crucial part
is not really simply a complaint about holding someone responsible for
that which is beyond his control. The complaint is not so much that one
reacts differently to the difference made by chance but that the kind of
reactions we have cannot be justified. Faced with the problem of
justifying hostile reactions we have sought suitable metaphysical
conditions which have produced an incoherent account of human
agency.
‘At this point many will be concerned that, even though we may have
found suitable reactions to wrong-doing which do not create problems
of moral luck, something important appears to have been sacrificed:
justice, Surely, it will be said, to abandon punishment would be to
abandon an important way in which we are able to react to evil, and to
abandon the practice would be to abandon the idea that justice must be
done. I will briefly suggest how our need for justice might be met
through requirements which do not involve punishment and which do
not create problems of moral luck.
Itis commonly held that a just punishment is one that fits the crime.
‘The idea that the punishment should be appropriately connected to the
crime does point to something important in our thought. We ought not
simply to do, nothing in the face of wrong-doing (whether a crime or
not) and we ought to concern ourselves with the victim's needs and
with what it is that the victim and society should want from the wrong-
doer. Furthermore, what is required from him should be in some way
related to his past misdeed. It may be thought here that any action
which falls short of punishment will amount to letting the wrong-doer
get away with what he has done. However, I suggest that the only kind
of suffering that truly fits the crime is remorse which arises out of the
wrong-doer’s realization of the wrongness of his deed.
Many criminals have served their sentences without feeling any
remorse for what they have done, and some will, without ever feeling
remorse, feel that they have paid for their crimes, But can wrong-doings
be paid for in this way? Is there not an important sense in which, if
someone is never brought to see the wrongness of his deed, then he has
‘got away with it’ in a way which is bad for his victim, society and for
himself. The attitude of the perpetrator of a wrong can be very
important to the victim. There is a big difference between the case352 BRYNMOR BROWNE
where a wrong has been done and the wrong-doer does not care about,
or even seems to enjoy reflecting on, his misdeed and the case where the
wrong-doer feels genuine remorse for his deed. There is, of course, a
kind of suffering through which a person who feels remorse must go,
and it seems to me that this kind of suffering is both appropriate and
good, It is good for the wrong-doer, in the sense that he would not wish
to be without it and its presence in him may also help the victim,
The remorse which the wrong-doer eels is appropriate because it is
internally connected to his past misdeed, rather than contingently
connected, as is the suffering of the prisoner. The cause of suffering in
the case of remorse is also the intentional object of remorse. The
suffering which is involved in feelings of remorse is an expression of the
realization of the wrongness of the deed, It is at the same time part of
what it is to realize the wrongness of the deed; we would not say of a
person who felt no remorse for a serious wrong-doing that he truly saw
that he had acted wrongly.
T have suggested that wrong-doings cannot be paid for and that real
justice requires that the wrong-doer feels remorse. What is more,
although a person who feels remorse suffers, he suffers in the right kind
of way, a way which involves no problem of unfair treatment and hence
no problem of moral luck. Of course, there will be some cases where a
person shows no remorse. Here I am only suggesting what [ believe is a
fundamental moral requirement for real justice, However, it should be
noted that no system of justice guarantees the required results in all
cases
Even though the law recognizes, and allows for, extenuating
circumstances in certain cases, the view of responsibility which lies
behind considerations of this kind is one which attaches responsibility
to individuals’ actions in such a way that the agents are regarded no
differently than complete instigators would be. But it is not only chance
events which contribute to our actions. Most of our moral life is a life
shared with others and others affect our character and the situations in
which we act: in fact, it is others who help to create the framework for
our actions. Each of us is influenced positively or negatively by what
others do or fail to do. Just as they share in our actions, so we share in
theirs. Once we see that contingent events play a part in all ofan agent's
actions we must also concede that most of the contingent elements
which help to form an agent’s acts are present through the actions of
other agents. In this way agents can be said to share in each other's
actions
When someone becomes bitter, and eventually acts wrongly, as a
result of being treated badly by others, those ‘others’ share in his action,A SOLUTION TO THE PROBLEM OF MORAL LUCK 353
however remotely. Of course, chance factors will also play a part in the
actions of those people who caused the person to become bitter and no
doubt the bitter person will, along with the others, share in the actions
of those who are affected by his bitterness. It is partly because we have
had an overriding concern to punish that we have held a certain view of
responsibility and have been less able to see that we are responsible in
many different ways for what others do. Once we no longer need to ask
about responsibility with a view to punishment it will seem less strange
to accept that there is no one single basis for attributing responsibility
and that we all share in each other’s actions.
‘An agent’s history always helps to form his actions. In many
situations one feels oneself to be totally responsible for what one does,
and as a difference-maker one is, in a sense, in that position, But this is
not an entirely true perspective from which actions can be judged. It
presents an incomplete picture, though one which may, in practice, too
frequently form the basis of the way that we tend to judge others.
From a first-person point of view we stand, at the moment of acting,
ata place from which we rarely see the influence on us of past events.
We act, possessing a certain kind of character and certain values,
which, although they are relevant to what we will do and to how we will
see our situation, will not seem to play a part in our action. But this is
also a misleading perspective from which to judge others. When there
are riots in the ghettos and vicious actions are carried out, our
incomprehension is based loosely on a picture of me now, being there then.
‘How could he do that?” we hear. The answer is that many actions are
comprehensible with the history and perspective of the agent who
performs them but not without it
When someone does a wrong, our attention is focused upon them and
we blame them for what they have done, They stand alone, without
their past influences or many of the circumstances which led to the
action. When we react to them we treat them as if they were the sole
cause of the wrong. Of course, they are responsible in a way that others
who may have influenced them are not and in a way that chance cannot
be, but what they will be responsible for should be viewed as
contributing, through their decision or action, to what happened.
Criminals are treated as complete instigators and are generally held
to be totally responsible for their wrong-doings. I believe that our
attitude to the offender is deeply imbued with scapegoating. Our
attitude towards the wrong-doer is rather like it would be towards a
totem which represents entire responsibility for the wrong. He stands in
for all the other contributors, taking on their sins (contributions), ready
to be crucified, thus absolving all.354 BRYNMOR BROWNE
Tt may be argued that we do not hold people totally responsible in
this way. I offer three considerations in support of my claim that we do
hold agents completely responsible.
1. We treat them harshly.
2. Wedo not look around for those who may have contributed to the
agent’s character, except in exceptional circumstances, where
their influence has been immediate or overwhelming and helps to
‘mitigate’ the offence. It is no part of our idea of justice that we
should seek out all those who have helped to create the conditions
which lead to the wrong-doing. In any case, itis not as if we seek
these others with a view to administering lesser sentences for their
contributions.
3. We condemn the man as well as punish him for what he does,
particularly in the case of serious crime. Allied to this is the
tendency toidentify a person’s whole being with his wrong action.
I have argued that if our attitudes were different there would be no
friction between our need to react to what the agent has done and our
need to treat him fairly. Once we abandon certain attitudes and
practices such as punishment, we are able to change our approach to
responsibility. Without certain of our present attitudes and practices
(punishment) there would be no problem of moral luck, even though
there would be many cases where we could speak of agents being
morally lucky or unlucky, such as the drunken driver who nearly kills
someone. In such cases it is still proper to make the observation that the
agent is morally lucky; but if correct attitudes to wrong-doing are in
place there will be no problem of moral luck, which is essentially a matter
of fairness in our treatment of agents.
So far I have restricted my discussion to evil actions and to actions
with bad consequences. However, luck comes into play where good
actions are involved, and people can be morally lucky in these contexts
too. This fact may at first seem to be at odds with my analysis, as my
main contention has been that the cause of the moral luck problem lies
in our wrong attitudes and that hostile attitudes directed against the
agent who is morally unlucky are at the core of the problem. It might
seem on the face of it that a problem of moral luck should not arise, for
example, in the context of rewarding. However, I shall argue that what
we find in cases of rewarding and praising actually supports my view
that the problem of moral luck is caused by our attitudes
Where rewards are offered as a means of encouraging others to do
something, for example, to find a lost dog, a minor problem of moral luckA SOLUTION TO THE PROBLEM OF MORAL LUCK 355
arises if, for example, all look equally hard, but one is lucky and finds
the dog. The question arises, “Are they not all equally deserving”;
surely we cannot reward someone on the basis of a difference that
chance has made. My position is that in such cases there is a problem of
fairness and it arises out of the instrumental attitudes involved. By
offering rewards we appeal to considerations which are extrinsic to the
kind of morally good motivation which many would show in searching
for the dog and one detracts from the true value of their actions.
Of course, rewards are not always offered as a means of getting
people to do something; at times, people may offer an unexpected
reward after someone has done a good deed. We should note that on
such occasions some are offended by being made such an offer which
appears to put the action for which they are being thanked on a
different footing. It may be very important that people are rightly
motivated towards each other, and accepting or offering a reward may
take something away from the action for which one is being rewarded.
Sometimes a person may feel deeply that he wants to express his
thanks to someone with a gift. Surely there can be no objection to this.
Indeed, I have none. But does not this action of expressing one’s
gratitude to the one who finds the dog create a minor problem of moral
luck? Is it fair that the finder should be thanked differently from the
others who looked (I take it that they were thanked too). My reply to
this suggestion is that there is nothing wrong with giving as an
expression of thanks and that there is no problem of moral luck in this
case. Should anyone perceive this situation as one which gives rise to a
problem of fairness, then I suggest that it is the attitude of the person
who thinks that there is a problem which is at the heart of the difficulty.
It is probable that such a ‘problem’ could only manifest itself against
the background of practices such as punishment.
A person who saves a life will be praised more for his action than one
who merely attempts to save a life. Again, I would say that thisis a case
of moral luck which is not a problem, for the same reasons as those
outlined above, but additionally, I would suggest that there also seems
less of a problem where praise is concerned. The person who tries to
save someone and fails cannot be treated in just the same way as the one
who succeeds, nor will there be any real temptation to say that he should
be. In Nagel’s case of the murder and the attempted murder there is a
considerable initial anxiety about blaming agents for what is
contingent to their wills; but in the case of praise the initial problem
seems less worrying. This is because the attitudes which are involved in
praising are not against the agent and are not, in themselves, wrong.356 BRYNMOR BROWNE
We praise the scorer of the winning goal more than the player who
makes a more worthy attempt and fails. I agree that we might see
something odd in praising where chance plays an obvious and crucial
role, but I have not denied that control plays an important part in
appraisal. Praising the footballer for a lucky goal is not an odd thing to
do, but failing to see that the player whose skilled efforts did not lead to
a goal is a good player who was unlucky would be (o miss something
important.
Tt seems to me that if we see a problem in the case of praise, then it is
because our attitude to praise is itself wrong. Praising can be viewed asa
kind of celebration in which we all share, or it can be viewed as a
comparative judgemental exercise which sets person against person in
moral competition. If we view praising in this latter way, then our
attitude seems rather like that of those who regard another's good
fortune as their own loss. We worry about the distribution of praise as if
points for the opposition were our relative loss. However, if'we do not
regard matters in this way, asit seems to me we should not, then there is
no problem of unfairness in praising persons differently for their
suecesses and failures.
Thus, we have a serious problem of moral luck where punishment is
concerned and where hostile reactions are involved, a less serious
problem where rewarding is involved and no problem where praise and
gratitude are concerned, which is what we would expect to find if my
solution is correct. The solution in all these cases is to be found through
modifying our attitudes and practices in line with deep-seated
intuitions which have for so long made many feel uncomfortable with
practices such as punishment and many of its associated attitudes, the
attempted justifications for which have led us into incoherence.’
St David's College, Lampeter
University of Wales
I thank Prolessor Robert Sharpe and other members of staff at St David's
University College, Lampeter, for their help and encouragement over the years, and in
particular to Dr David Cockburn for his time and invaluable comments on my work