Foot, Philippa , Griffin Professor of Philosophy Emeritus at the University of California, Los Angeles, and an Honorary Fellow

Philippa Foot

The title of this lecture will, I am afraid, have suggested many fascinating topics
about which nothing is to be said. I am not going to argue, with G. E. Moore, that
aesthetic experience is an intrinsic good which ought to be pursued for its own sake.
Nor shall I join in the debate as to whether moral considerations are relevant when
we judge works of art. I shall talk about moral and aesthetic judgement, but in order
not so much to relate as to contrast them, and that from a special point of view. What
I want to do is to follow a hunch, originally a mere suspicion, hard even to formulate,
that there is some element of fiction and strain in what we say about right and wrong,
while our appraisal of aesthetic objects is relatively free from pretence. How far
others share this uneasiness about morality I do not know but I am certainly not the
only one to have such thoughts or to find this contrast. One often hears those not
conditioned to suppress all thought of the mere possibility of an attack on morality
say that they are sceptical about moral judgements, and it may be a successful
tactical move, always supposing that one wants to ride them off the sceptical
position, to ask whether they are also sceptical about the merits of works of art. My
present intention is to pursue the matter rather than to close down the debate. I shall
argue that the original, half-formed, thought about morality, and the contrast between
moral and aesthetic judgement, has something behind it. If I am right we do maintain
certain fictions about morality, and they are even reflected in the forms of language

end p.5

that we use, while there is no comparable lack of candour in what we say and think
about works of art.
I must begin by raising, yet once more, the old question as to whether moral
judgements are subjective or objective. That they are, simply, subjective seems to
me certainly false. For to impute subjectivity implies a comparison with the words in
which a man expresses or reports his feelings, sensations, aims, allegiances, and so
on. Here questions of truth or falsity either do not arise at all or else arise only in a
special way. Where we have the verbal form of an exclamation, as when someone
says 'alas!' or 'ow!' or 'how nice!' we cannot tell him that what he said was false,
though we can suggest that he is pretending or being insincere. Where he uses a
declarative sentence, as in saying 'I am dismayed'or 'that hurts'or 'I like that', we may
say 'that is not true', but only if we think he is lying or at least insincere. Whichever
form of words he chooses there is no room for the kind of mistake that he can make
when his eyes have deceived him, or when he has drawn the wrong conclusion from
some piece of evidence or other, which might of course, in a special case, be
evidence about himself.
That all moral judgements are subjective in this sense is ruled out, rather trivially, by
the fact that some, at least, are not asserted directly but rather derived; this at once
breaks the analogy with subjective utterances by introducing a possibility of error not
due to insincerity. Nothing follows, however, about the premises from which such
conclusions are drawn, and we are not yet committed to any view about what their
status may be. Now about this there has been a great deal of controversy, some
philosophers arguing that a man can choose for himself, so long as he meets formal
requirements of generality and consistency, what his ultimate moral principles are to
be; while others insist that certain criteria of good and evil belong to the concept of
morality itself. The first, or formalist, position seems to me indefensible, implying as it
does that we might recognize as a moral system some entirely pointless set of
prohibitions or taboos, on activities such as clapping one's hands, not even thought
as harmful, aggressive, treacherous, cowardly by the community in which the

prohibitions exist. A moral system seems necessarily to be one aimed at removing
particular dangers and securing

end p.6

certain benefits, and it would follow that some things do and some do not count as
objections to a line of conduct from a moral point of view. There may be strict proof of
some moral propositions, such as that Hitler's treatment of the Jews was morally
indefensible. Many philosophers have rejected this possibility on the general
theoretical grounds of a logical gap between 'is'and 'ought'. I shall not here renew the
arguments against this supposed refutation. It seems to me that there is no difficulty
in principle in making such a derivation, and that there are, as I said, starting-points
fixed by the concept of morality. We might call them 'definitional criteria' of moral
good and evil, so long as it is clear that they belong to the concept of morality—to the
definition and not to some definition which a man can choose for himself. What we
say about such definitional criteria will be objectively true or false.
So far we have been describing elements of objectivity in a moral system, but there is
still some room for subjectivity for all that has yet been said. For even if the
arguments against fixed starting-points in ethics are invalid, and even if there actually
are definitional criteria of moral good and evil, allowing some moral propositions to be
proved from the facts, it does not follow that we can settle all moral questions in this
way. There could be both fixed starting-points and an element of 'play' in the system,
allowing different and irreconcilable points of view about certain things. I think that
this is probably the actual position, and that the concept of morality while it fixes a
great deal also leaves quite a lot open. It seems, for instance, that while one can
determine from the concept of morality that there is an objection to murder one
cannot determine completely what will count as murder. Thinking about the problem
of abortion I come to the conclusion that there is a genuine choice as to whether or
not to count as a human being, with the rights of a human being, what would become
a human being but is not yet capable of independent life.

It would be quite another matter to suggest that one could put the distinction between what is and is

not a human being where one chose, and still have an interpretation of the principle that there is a
moral objection to the killing of human beings.

Again there seems to be a more general choice of starting-point involved in the
decision as to how far we are going to protect an individual against the claims of the
community,

end p.7

where the interests of the majority clash with his. It is extremely difficult to find a clear
and obvious principle when we start looking into the problems, e.g., of medical ethics.
We ourselves have a strong objection to the idea of using one person for the benefit
of others, and it probably guides our intuitions in many cases. It does not seem clear,
however, that one could rule out of court the principles of a strict utilitarian who
would, at least if he were consistent, allow things that we will not allow in the interests
of cancer research. And so it may be that at some points we really do find the kind of
ultimate breakdown in moral argument that has been suggested as a quite general
possibility by those who refuse to admit definitional criteria of right and wrong.
It seems, therefore, that philosophers on both sides of this line of controversy must
ask themselves about the status of what we might call 'contingent principles'used in
forming moral judgements. My contention is that if we really reach such starting-
points in a man's morality we have something that can only be stated subjectively. By
this I mean that the words will operate subjectively however much we insist on an
objective form. Let us see what the arguments against subjectivity are worth at this
point.

It is often said that no moral utterance could simply state a fact about one's own
attitudes, the objection being that when one person says 'X is permissible' and
another 'wrong' or 'not permissible' both statements cannot be true. But that, in a
discussion of subjectivity, is just the point at issue. If the declaration of a contingent
principle really has the characteristic given to it in the hypothesis, if, that is, it really is
a starting-point, why not say that the only truth of which it is capable is that
(depending only on veracity and sincerity) which can belong to a subjective report
such as 'that hurt' or 'I am hostile to that man'? No doubt it will be objected that this
cannot be right, because we actually use 'true' and 'false' in a different way in
connection with 'moral judgements', saying 'false' not to charge a man with lying or
insincerity but to state an opposed point of view. That we do this is, of course, true,
but the question is what the role of this piece of usage is once a statement of a
contingent principle has been reached. It is certainly not necessary in order to allow
us to speak of the man who says 'p' and the one who says 'not p' as disagreeing.

end p.8

'Disagree' is (to use an expression of Miss Anscombe's applied by her in a different
context) 'a light word'. If you find something pleasant and I do not, or you find some
food delicious and I do not, we can say 'how we disagree'. My suspicion is that the
existing use of 'true' and 'false', and the choice of an objective form of expression ('it
is right'), does have a role but a rather disreputable role. When we say that
something 'just is' right or wrong we want to give the impression of some kind of fact
or authority standing behind our words, though by hypothesis both are here ruled out,
maintaining the trappings of objectivity though the substance is not there. Perhaps
there is not, in the language already, a subjective form of words which will say just
what we want, but we do not have to keep the language as it is.

Anyone who suggested that we could use the existing expression 'think it right' would be very far out

indeed. For this implies just the contrast between truth and opinion that is being denied.

This, then, is the first place where we see a statement of moral position dressed up
to look like something other than it is. Do we find the same gap between appearance
and reality at a similar point in our appraisal of works of art?
There is, of course, a great deal of difference between the activity of passing moral
judgement and that of aesthetic appraisal and it is perhaps strange that philosophers
have often tried to find a theory to cover them both. In the first place Nelson
Goodman is surely right in insisting that the decision about what is good and bad
does not play an all-important part in the critic's work:'works of art are not race-
horses, and picking a winner is not the primary goal'.

Languages of Art (Indianapolis, 1976), 262.

Secondly, the picture of an ultimate breakdown in argument does not have the same
hold on us in aesthetics as it does in ethics. This is partly because a critic tries to
display the characteristics of his subject but rarely to offer proof that it is good or bad.
Moreover, description does not present the object of art as it presents the object of
moral judgement, so that it is not obvious even what would be meant by saying, of a
work of art, that two people might agree on all the facts and still disagree about the
values. The idea of 'all the relevant facts' seems to have no application if we are
thinking about works of art. Nor

end p.9

is it clear that there could be contingent principles in aesthetics as in morals. This
may sound surprising, for, of course, there have been many conflicting
announcements about what makes a work of art 'good' or 'great'. One has only to
think of Tolstoy's insistence that good or true art must produce a feeling, quite distinct
from all other feelings, of 'joy and spiritual union' with the author and with others who

appreciate the work. Such an announcement does not, however, have the status of
our 'contingent principles' in ethics. For there is a touchstone by which Tolstoy's
judgement may be tested, and by which, incidentally, it may be shown to be
insufficient. Someone may, if he chooses, restrict his own attention to works of art
that satisfy certain criteria. What is great or good is determined, however, by what
people find in certain objects, that is deeply interesting and significant to them. If
someone insists that what is thought of as a defect is really a merit he says
something that relates necessarily, though of course in no simple way, to the
reactions of others. He has the alternative ready to hand in an overtly subjective
statement of his own appreciation or taste. In matters of moral judgement it is hard to
accept subjectivity where it genuinely exists. It seems far easier in aesthetics, and
this may be because it is not our own conduct that is in question. We are apt to be
haunted by the thought of authority when it is a question of what we are to do, and
the thought that somehow and somewhere it is written down is very compelling
indeed. It is not very tempting to refer aesthetic values to the will of God.
We find, I think, the same pattern if we raise the question of the relativity of
judgements in morals and aesthetics. I will first try to show that we are over-resistant
to theories of ethics that allow the relativity of some moral judgements. Relativism is
often these days said to be absurd as a theory of moral judgement. As a theory of all
moral judgement it is, of course, false if the thesis of definitional criteria is correct.
(There would be nothing relative about the truth of the proposition that Hitler was a
moral monster.) But if what we have also said about contingent, alternative, starting-
points for a moral system is true it should follow that there are moral judgements for
which a relativistic account will be right.
Since there is a good deal of confusion about what is meant by a relativistic theory of
ethics, we should discuss this before going on.

end p.10

Sometimes, for instance, relativism is contrasted with absolutism, so that a relativistic
theory says that circumstances may always alter cases while its rival holds that
certain kinds of action are always and everywhere good or bad. So, to deny
relativism in this sense, which is not the one in which I am interested, one would
need to say for example that if the description 'torture' or 'judicial condemnation of the
innocent' applied to an action it was, in any circumstances, morally inadmissible. I
mention this distinction only to set it aside. It belongs to our present topic only
indirectly in providing other examples of what might be contingent principles.
Perhaps it will be best to begin not with a definition of relativism but with some
examples of judgements outside ethics for which a relativistic account would be
correct. Many could be taken from the province of taste, as when, for example,
clothes are said to be elegant or smart. Again we might consider the statement that
someone is good-looking or fair of face. Other examples could come from
gastronomy. The food at a certain restaurant is said to be good; or it is said that a
particular process, say burying eggs for long periods, or letting yak's butter go rancid
before floating it on the tea, gives the product a delicious flavour. Presumably no one
will deny that the truth of such judgements is relative, not, of course, to the speaker's
tastes, since one may make a mistake in thinking one's wife good-looking or the
flavour of one's cooking good, but rather to the reactions of certain people at certain
places and times. This is not to suggest that there is some simple connection
between, say, the preferences of a majority and the judgement of taste; to give a
proper account one would have to look at each case separately and ask, for
instance, whether connoisseurs are recognized in the given area, and if so what kind

of authority they are allowed. The relevant point for this discussion is that reactions
do vary a great deal from time to time and place to place, and that while sentences
predicating such things as handsomeness or good flavour have a truth value as
uttered in a given context they cannot be used 'across cultures' as it were, since it is
impossible to extract the utterance from both contexts and ask questions about the
truth. It would quite obviously be ridiculous for us to say that our opinion about the
Tibetans' tea is correct and theirs mistaken, nor would we

end p.11

necessarily dispute questions of female beauty even with a traveller from the
eighteenth century, never mind with a representative of the ancient Maya peoples
who practised cranial deformation and on purpose made their children cross-eyed. It
is not that such judgements do not have a truth value, but they do not have one
unless relativized in a particular way.
Another example, which may be more controversial, is to be found in the use of the
sentence 'that's a good thing'. Moral philosophers have sometimes spoken of this
proposition in a portentous way, as if it passed judgement on the state of the
universe from some lofty point of view. And indeed it may do so, if the speaker has
lofty interests, and knows or supposes that his hearer has them too. But 'It's a good
thing that such and such' may be said by the robbers as well as the cops, and by
those interested only in making money on the next race and getting safely home. It
would be ridiculous to say that when the backers of rival horses reacting to some
news affecting their prospects, say 'a good thing', 'a bad thing' the truth of what they
said is to be judged by the interests of racing, never mind the welfare of mankind; nor
can a third speaker say 'which of them is right'. Yet this type of utterance is once
more to be contrasted with subjective statements such as 'I am glad'. One may make
a mistake in thinking it a good thing that such and such has happened, and two
speakers who have the same desires and allegiances may challenge each other's
opinion about whether it is.
So far all the examples of truths relative to a context of utterance have been from the
class of evaluative judgements, but there are others from elsewhere. We might think
here of the sentence 'N.M. is tall'. For suppose that our man, N.M., being about 5 ft. 9
in., goes to live first in southern Italy and then in California; he will properly be called
first tall and then not tall without any change in his height, while in mid-Atlantic 'N.M.
is tall' may have no use at all.
It will be apparent that the relativity in which we are interested is a special case of
that belonging to any sentence containing indexical expressions such as 'I' or 'here',
or tensed verbs, since these too are true as uttered in one context but not in another.
It is not easy to give an account of the distinctive mark of relativity as we want to
consider it in ethics.

end p.12

Something must be said about this, however, since one would not call a theory of
ethics relativistic in that it allowed for an element of relativization coming merely from
the use of indexical expressions in making moral judgements. If we speak of
relativism in ethics we mean something more than this. Roughly we might say that
the relativity in which we are especially interested is truth relative to standards or
tests. So we know that when 'he is tall' is said in California the height of Californians
must be appealed to; when an Eskimo speaker talks to an Eskimo audience about
good food what he says is to be tested by Eskimo tastes.
Since the context of utterance is necessary to determine the truth or falsity of these
sentences for which relativization is necessary there is, of course, no question of
contradiction between speakers saying 'p' and 'not p' in relevantly different contexts.

The meaning of 'p' may nevertheless be the same, though it is a difficult problem as
to when we will say that this is so. It seems unproblematic in the case of sentences
containing indexical words (such as 'I am English') and also for the case of being tall,
since we would be inclined to gloss 'N.M. is tall' as meaning, in all contexts, 'N.M. is
well above average height' rather than in one context one thing and in another
context another. What of the sentence 'the flavour is good' as uttered by people of
different gastronomic backgrounds? One would say that the words do have the same
meaning, especially if the relation of the judgement to preferences were exactly the
same in the two communities. It would be harder to know what to say if in one
community but not the other there were connoisseurs of flavour with a position like
that of our connoisseurs of wine. Luckily this problem does not have to be solved just
now, since it is only where 'p' is thought to have a constant meaning that there will be
any question of contradiction between 'p' and 'not p'. What is important to us is that a
difference of context of utterance will often remove apparent contradiction even here.
Let us now consider whether a relativistic account should be given of moral
language. It was pointed out earlier that a completely relativistic account is out of the
question for those of us who believe that there are some criteria of good and evil
implied in the concept of morality. Moreover, no moral judgement will depend for its
truth solely on its utterance

end p.13

in a given context, since all are subject to the definitional criteria. Nevertheless, if it is
the case that different speakers can validly relate their conclusions to different
contingent principles there is an element of relativity in morality. Perhaps a good
model for comparison would be that of a discussion between two or more people
about what it would be best to do where certain aims were implied in the question
that had been raised, as e.g., what would be best for the college as discussed in a
college meeting, but where some other aims and principles might nevertheless not
be shared. In such a case we would talk together about what was best until some
divergence of principle actually appeared, though after that the discussion might
have to end. This seems to be how we do talk about morality with those whose
principles resemble our own.
I think it likely that even this modest amount of relativity will not easily be accepted
where ethics is concerned. For if it is accepted we must admit that in certain cases 'it
is wrong to do X' will be true as said by A who has one set of contingent principles,
and false as said by B who has another, the description and circumstances of X
being taken as the same for A and B. And we do not like to admit this, particularly
when thinking of ourselves as A or B, as if we felt that the word 'true' in the sentence
'As uttered by him p is true' somehow implied a weakening of our own allegiance to
principles from which not p can be concluded. Yet we have to say that a Tibetan
uttering a sentence meaning 'Rancid yak's butter gives a good flavour to tea' would
be saying something true, and that we could find similar examples having to do with
elegance or personal beauty. As said by some other person the sentence might be
true though as said by us it would be false. No compromise of our own position is
involved, and if we refused to apply the word 'true' here this could only support the
fiction

of a superior point of view. This is, of course, exactly the point about the moral

divergencies we have been considering, i.e. the ones stemming from the adoption of
different contingent principles. So long as we still have something to bring forward to
show why particular principles should not be adopted we have a genuine possibility
of showing that ours is a superior point of view, and it is very important that nothing

that can really be said should be squeezed out. But if someone merely asserts that
something is right or wrong, as for instance he may

end p.14

say when discussing euthanasia 'to kill a human being is never permissible', he
should admit the relativity of judgements based on this contingent principle. If his
allegiance to it really requires a fiction about its status there is something wrong, and
it may be that some people who do not mean to base their morality on religion would
change their moral views on such matters if they no longer covertly referred to an
authority such as God. On the other hand some contingent principles would surely
survive. People do care, for example, that individuals should not in everything be
sacrificed for the good of others, or that the unborn should be counted as in the
human community. They are ready to fight for such things; why should they not
continue to do so?
When thinking about art we do not, it seems to me, have the same worry about
relativism as we do when thinking about morals. Or at least we do not have the same
resistance to taking things as they come. If we actually have something to say in
criticism of the art of some other time or some other culture, as that it is sentimental,
we expect to be able to show that this is so, and to be backed up by the reaction of
other generations as well as by our contemporaries. If there were genuinely nothing
to say and nothing to show, as we suppose may be the case at certain points in
morals, we would be most unlikely to insist that somehow, nevertheless, we must be
right. In fact, as everyone knows, there is a most surprising, and rather moving,
agreement between peoples whose civilizations are completely different, and who
may even be culturally isolated from each other. It was not to be taken for granted for
instance that we would recognize the expressiveness of figures made by Han
craftsmen, or in pre-Columbian Mexico, or that this would have the importance for us
that it does. Sometimes recognition or appreciation fails. But then we are increasingly
likely to think of the matter like that, and not to condemn what we do not understand
or like. Obviously this has something to do with the fact that we are, psychologically
speaking, much freer from anxiety in relation to art than to morality, and also that our
thoughts about artistic merit are not haunted by a historical connection with religion.
Moral judgements regulate our conduct in just those areas which arouse the deepest
feelings of guilt, so that we want to erect the strongest possible barriers against what
we fear we might do;

end p.15

aesthetic judgements guide our conduct in relatively calm waters when they guide it
at all. Thus we are not nearly so likely to speak and feel as if things have been laid
down for us in spite of the overt belief that they have not.
These considerations are no doubt connected with the final contrast that I shall draw
between the way we think about morality and the way we think about art. In the first
case, I want to say, we maintain a fiction of inescapability, while in the second we do
not. In a sense, of course, it is right to think of moral judgement as inescapable. A
man does not cease to be cruel or dishonest because he does not mind if he is cruel
or dishonest, and these characteristics may also wreck his life. But suppose that he
does not care, and moreover argues (it just might be truly) that in his position
wickedness is likely to bring him more benefit than virtue. He is a ruthless amoral
man, and moreover in such a special situation that we are not able to insist that even
from the point of view of his own good he has reason to abandon his cruelty and
dishonesty. We ask him if he does not care for example about the suffering his action
will bring to others, and he says, and we believe him, that he does not care. What
can we say to such a man? What we do say is that he should care, and that he

should not

do the things he does; we must, however, look carefully at this use of

'should', and in particular ask whether it implies that he has reason to change his life.
Normally 'should' does carry a necessary connection with reasons for acting, since
we use the word, for example, in giving advice, and advice must relate to what the
agent wants (for himself or others) or to what he will care about at some future time.

I am not here committing myself to a simpliste view of advising. As a corrective to this, see B. J.

Diggs, 'A Technical Ought', Mind (1960).

We also use 'should' in contexts such as that of a college meeting, where common
aims and principles are presupposed, and so far the use of 'should' is parallel to that
of 'a good thing'. It is different, however, in that 'should' is not here applicable to
someone not sharing in the common enterprise. If, for instance, it is important for the
college that a certain shopkeeper moves his business we cannot on these grounds
say that he should move, although we do say that it would be a good thing if he did. If
we are to say that he should move, then, unless we are urging some duty upon him,
we

end p.16

must bring considerations related to his desires, as for instance the lack of custom
where he is. I am not, it should be stressed, saying that these non-moral uses of
'should', carrying a necessary connection with reasons for acting, must relate a
course of action to something by which the agent will be moved. For it is often the
case that one should do something, like getting up in the morning, for the sake of
something one does want (and it is true now that one wants it) as for instance to be a
good philosopher, but which, at this moment leaves one cold. (Compare being fond
of someone though just at the moment one feels one never wants to see him again.)
Moreover the 'should' of advice can properly be backed up by some consideration
about what the subject will want even if there is no sense in which he wants it now,
and this too is enough to give reasons for acting. It is a clear case of irrationality if a
man takes no account of the future penalty of his present action. When backed by
moral considerations 'should' is used differently, and as we noticed earlier a man may
be told that he should do something required by morality without any connection
being traced between such action and his present or future desires. So though in
some ways we may think of the 'should' of morality as comparable to any other, as if
members of the community were meeting together with the particular aim of making
good arrangements for a common life, the analogy breaks down where a man who
explicitly puts himself outside the common enterprise is still told what on moral
grounds he should do. That this is what we say is among the data, and it is obvious
why we have this usage in the language. Firstly we do not expect people to stand
outside the moral community as does our ruthless amoral man, and secondly we
want to be able to influence those who do. What is not, however, clear is why they
should do what we want. Someone who is a lover of justice, and who cares about the
good of other people, has the best of reasons for taking account of morality. Can we
say the same of our uncaring man? If he does what we say he should do what will his
motivation be? Will he not simply be afraid of our disapproval, or following a training
which he is too timid or conventional to throw off?
I am sure it will be replied that the 'should' of morality necessarily gives reasons for
acting to any man, but this is what I want to question.

end p.17

Why should we think that the connection with reasons exists necessarily even in the
case of a 'should' that works, like this? In one sense it is true that moral judgements
necessarily give reasons for acting, but in that sense it is also true that, for example,
codes of etiquette must do so. (We would not speak of a code of etiquette unless

people in general guided their conduct by such rules.) Yet a man may ask whether
he has any reason to obey some other code that he has been taught. Why should he
not ask this also about the moral code, claiming to free himself from morality as many
have freed themselves from codes of manners? It is no use our reiterating the word
'should' and 'shouldn't', since this piece of usage is not being questioned, only the
connection with reasons for acting when 'should' is used like this. If we want to use
the moral 'should' as we do in applying it even to a man indifferent to the ends of
morality must we not accept the penalty that the connection between what a man
should do and what he has reason to do breaks down? What we are inclined to do is
to ignore all this, insisting on the word 'reason' without being able to show why we
should use it in the case of morality though not, like this, in the case of etiquette. The
rationale is lacking though the motive is clear. We are, naturally, concerned about
someone who doesn't care what happens to other people, and we want to convict
him of irrationality, thinking that he will mind about that. Outside moral philosophy we
would not think of the cool and prudent, though wicked, man as specifically irrational
in his conduct; outside philosophy we also know that there is nothing one can do with
a ruthless amoral man except to prevent him from doing too much damage. To say
that since his conduct is immoral we can tell him of some reason why he should
change it, or that he necessarily has reason to alter his ways, seems yet another
case of keeping up a pretence. We speak as if there were an authority in the
background to guarantee that wickedness is necessarily foolishness, though the
'binding force' of morality is supposed to be independent of such an appeal. Would it
not be more honest either to change the language or else to recognize that the
'should' of moral judgement is sometimes merely an instrument by which we (for our
own very good reasons) try to impose a rule of conduct even on the uncaring man?

end p.18

Once again, when we turn to aesthetics we find that the same problem does not
arise, and this is connected with a fundamental difference in the judgement of art and
of conduct. If we ask, as Plato might have done, what good is rendered by good
conduct, and again by good art, and to whom it is rendered in each case, the
answers will not be the same. The good of good action, while it may indeed accrue to
the man who does the action, will very often go rather to another. A man who is just
or honest or charitable produces a benefit for others rather than for himself, which is
why we may not always be able to show that an unfeeling man, no lover of justice,
has reason to be moral whatever the situation he is in. The good of good art, on the
other hand, lies in such things as the pleasures of the imagination, and in general in
the interest and enjoyment that we get from works of art. And since the one who
chooses the book to read, the picture to look at, the music to hear, is usually the one
to whom that good will come, there is no difficulty in seeing that he has reason to
choose the good and reject what is bad. Where this connection breaks down, as it
does in the untypical case of someone choosing a work of art but not able to
appreciate it, we would not say that nevertheless he should choose what was good,
unless in the hope that he might later come to appreciate it. We would recognize it as
nonsense to say 'The fact that a work of art is a good work of art is itself a reason for
choosing it, and never mind whether you will get anything out of it or not.'
I have now drawn contrasts at three different points between moral judgement and
the judgement of works of art, in each case with a conclusion unfavourable to moral
judgement. I know that what has been said about the appreciation of works of art has
been inadequate, and I hope that if you reject my remarks on aesthetics you will
consider the discussion of moral judgement as standing on its own. My main interest

in these topics is from the point of view of moral philosophy rather than aesthetics,
and what I should most like to do is to open a debate which might lead us to consider
changes in the way we talk about what is morally good or bad. Why, after all, should
we take it for granted that the form of language already developed is the one we
want?

end p.19

Moral

Relativism

Philippa Foot

Some philosophical questions interest only philosophers: they would never occur to
the plain man, and if he hears of them he may very well think that those who spend
their time on philosophy must be a trifle mad. There are, however, other problems, no
less philosophical and just as important, that are apt to present themselves to any
enquiring mind. One does not have to be a philosopher by trade or training to have
doubts, for instance, about free will; and it has even struck many innocent of
philosophy that perhaps the world looks undetectably different to different persons,
one systematically seeing as red what the other sees as green. The thesis of moral
relativism is one of these natural philosophical thoughts. Very many students,
beginning philosophy, are sure that relativism is true; and although they are often
taken aback when reminded that it is, for example, common for members of our
materialistic society to criticize this society for its materialism they usually think that
some adjustments will save the theory. One might therefore expect that moral
relativism would be a central topic among those discussed in classes and in the
journals. Surprisingly, however, the truth has for long been quite otherwise. Many
recent books on moral philosophy ignore the problem or give it perfunctory treatment,
and it is only in the last two or three years that strong, interesting, articles have
begun to appear in print.

See, for instance, Bernard Williams, 'The Truth in Relativism', Proceedings of the Aristotelian

Society

, 75 (1974-5); Gilbert Harman, 'Moral Relativism Defended', Philosophical Review, 84 (1975);

and remarks in David Wiggins, 'Truth, Invention, and the Meaning of Life', Proceedings of the British
Academy

, 62 (1976).

end p.20

Why was the subject so long neglected? Probably it was because few of those
teaching, and writing, philosophy believed moral relativism to be anything they need
worry about: some thought they knew how to discredit it in a few easy moves, and
others supposed vaguely that it had been done. In fact, as we shall see, there were
elements in the prevailing theories of ethics—in emotivism, for instance—that made it
seem difficult even to formulate relativism except in a version that was indeed easy to
refute.
C. L. Stevenson is one of the few influential moral philosophers of the past thirty or
forty years to have treated the subject of relativism at any length. He argued, in an
essay called 'Relativism',

Ch. 5 in his Facts and Values (New Haven:Yale University Press, 1963).

that his own 'emotivist' theory of moral judgement gave a basis for the refutation of
moral relativism. Now what Stevenson actually says about relativism, what he takes it
to be, is rather odder than most people remember if they have not been reading him
lately: he says, for instance, that an account of moral judgement taking 'X is good' to
mean the same as 'I approve of X' is not relativistic, whereas the same theory would
be relativistic if requiring the speaker's name to be inserted instead of 'I'. The interest
of Stevenson's discussion is not, however, in these details but rather in a certain

assumption; namely the assumption that a relativistic theory identifies a moral
utterance with an assertion of psychological or sociological fact. The relativist is
supposed to identify the thought that a given action is morally good or bad with some
proposition about the reactions that people have to it; and it is on this that
Stevenson's 'refutation' depends. He points out that the identification must be
mistaken because the two types of propositions are backed up in different ways. If
someone is asked why he thinks certain things morally good or bad he does not set
out to show that some individual or group really does have this or that reaction to it,
but tries to bring forward facts about the action itself. Moreover, in putting forward his
moral views he expresses feelings or attitudes and tries to change the feelings and
attitudes of others, whereas a statement of psychological or sociological fact lacks
this dynamic aspect.

end p.21

One does not have to share Stevenson's emotivism to agree that moral judgements
are not descriptions of reactions, since his first argument seems sufficient for the
proof. Stevenson will, however, have refuted moral relativism only if his assumption
is true of all its versions. And it seems implausible to deny that other models are
possible. For outside ethics we actually find judgements that do appear to be
relativistic but not in Stevenson's sense.
It will be worth spending a little time considering what might be meant by calling
certain types of judgements relativistic; and what relativism amounts to in those
areas in which it seems to belong. I am thinking, for instance, of certain judgements
of 'taste', such as those asserting that some people but not others are good-looking,
that some food or drink is appetizing or delicious, or that certain colours go well
together for furnishings or clothes. Here, it seems, we find wide variations in
judgements between different cultures and different generations. One does not have
to go as far as ancient Mexico to find a set of faces that we find ugly while supposing
that they were once admired, and while we think Nureyev's a better-looking face than
Valentino's there was a time when the verdict would probably have gone the other
way. It is obvious that there is the same kind of disagreement about the palatability of
food and drink; and combinations of colours once declared deplorable are now
thought particularly good. The old rhyme said that 'blue and green should never be
seen,' and black and brown were once seen as colours that killed each other as we
might now say that navy blue kills black.
The reason why such judgements seem undoubtedly relativistic is not, of course, that
a wide variety of opinions exist, but rather that no one set of these opinions appears
to have any more claim to truth than any other. But there is a problem here. For if the
differences in the application of concepts such as 'good-looking' are as great as this,
why are we so confident that at different times or in different places the judgements
are about the same thing? This difficulty must be taken seriously, and may lead us to
cut down the number of judgements that we would count as certainly relativistic, even
in the area of 'taste'. Perhaps some kind of relativism is true of many other
judgements, but relativism is most

end p.22

obviously true where we need set no limit to the variations in the application of an
expression, or rather no limits to its application within a given domain. This condition
seems to be fulfilled for our examples, but it would not have been fulfilled had we
been operating with concepts such as prettiness, or even handsomeness. It makes
sense to speak of another society as thinking good-looking just the faces we think not
good-looking, but not as thinking pretty just the faces we think not pretty. The
examples most suitable for the present purposes are those that are rather general,

and this is why I suggested considering the good-looking, the good-tasting, and the
good combinations of colours.
Let us suppose then that there are in different communities divergent sets of such
judgements which we have no hope of reconciling, and that in this area we also have
no thought of distinguishing the opinions of one group of people as right and those of
the others as wrong. Shall we say that this is because the judgements describe
reactions such as admiration and liking, and that reactions vary from place to place
and time to time? This, which fits Stevenson's version of relativism, is not, in the area
we are now discussing, the truth. To say that someone is good-looking is not to say
that his looks are admired, any more than to say that someone is likeable is to say
that he is well-liked. No doubt it is true that the concept likeable depends on reactions
of liking. And no doubt it can operate as it does only on account of shared reactions
of liking. Shared reactions are also necessary if the language of a particular
community is to contain a word like 'elegant', or if it is to be possible to say in it that
certain colours go well together. But there is no reason to think that the judgements
describe the reactions. One might as well think that 'is red' means the same as
'seems red to most people', forgetting that when asked if an object is red we look at it
to see if it is red, and not in order to estimate the reaction that others will have to it.
That one does not describe one's own reaction of admiration in saying that someone
is good-looking is shown by the fact that one may admit a mistake. That one does not
describe the reactions of others in one's community is shown by the fact that one
may accuse them of mistake. Nor is this kind of language empty, the mere reiteration
of the expression of one's own reaction. There is room here for the idea of showing,
even if not of proving or demonstrating. An

end p.23

individual who makes some very idiosyncratic judgement may simply be ignored or
told that he is out of his mind. But he may say something that his fellows find
instructive, either with an explanation or without it. I do not want to attribute to any
particular type of judgement one jot of (local) objectivity that does not really belong to
it, or any method for bringing agreement that does not really go along with it. But
distinctions are there to be made. It will not do for one of us to say that Charles
Laughton got up as the Hunchback of Nôtre Dame presented the appearance of a
good-looking man; but Laughton's brother's suggestion that he was, in his own
person, good-looking was a surprising but possibly corrective idea.
What this discussion shows is that if a relativistic thesis is true of the judgements
under scrutiny it does not assert relativism as understood by Stevenson. And yet it is
certainly relativism. For the key concepts will work as they do work—with a kind of
objectivity and the attribution of truth—only where there are shared reactions. Once
this background is left behind it is impossible to speak of 'right' and 'wrong', 'mistaken'
and 'correct', as we commonly do, and therefore it is empty to say of the judgements
of another group whose reactions are very different from ours that their opinions are
wrong. Our own discussions of these matters of 'taste' implicitly invoke the standards
set by our paradigms and our way of going on from them, and here we can speak of
right and wrong. But if we are talking of the views of another society we shall speak
of what is true by their standards and by our standards, without the slightest thought
that our standards are 'correct'. If the ancient Mexicans admired the looks of
someone whose head had been flattened, a proposition not about this admiration
may have been true as spoken by them, though it is false as spoken by us.
We have, then, a version of relativism true of some judgements and not vulnerable to
a Stevensonian type of refutation. The question is whether moral assertions might

similarly admit only of relative truth. It will probably be objected that this is impossible
because moral judgements do not depend on local moral standards as our
judgements of taste were thought to depend on local standards of taste. The thought
behind this objection is that a challenge to a moral judgement or moral

end p.24

system can be made 'out in the open' as it were, with no agreed method, formal or
informal, for showing that the challenge is justified. The idea is that there can be
disagreement, with each party thinking the other mistaken, even if there is in principle
no way of settling the 'dispute'. It is therefore supposed that an individual can
challenge the views of his own society not just in the way that we think it possible for
someone to query some judgement of taste, but more radically. He is to be able to
say anything he likes about what is morally good and bad, so long as he is
consistent, and is to be taken seriously as a man of very eccentric taste would not
be. It follows also that the members of one society may similarly challenge the moral
view of another society. No common starting-point is necessary, and nothing to back
up an accusation of falsity or mistake.
Anyone taking this position will insist that moral assertions do not have merely
relative truth. Local standards are supposed to be irrelevant, and there is to be no
point at which a set of moral opinions inconsistent with one's own are to be admitted
to have just as much truth. This is an argument that we should examine carefully; it is
perhaps not as powerful as it looks.
The case against construing moral judgements as relativistic along the lines that fit
judgements of taste has been made to depend, it seems, on some points of linguistic
usage. It was thought crucial that we can say of the moral opinions of our own
society, or of some other society, that they are 'mistaken'or 'false', and this was
described in terms of a 'challenge'to moral views that differ from our own. The
question is whether 'challenge'is the right description if words such as 'true'and
'false'are used as they are supposed to be used here; and in general whether it is
important that these words are or are not so employed. It will be remembered that by
our hypothesis talk of truth and falsity was to go on even in the absence of any kind
of proving or showing, or any possibility of proving or showing, that one view rather
than the other was in fact true. And this is, of course, a situation very different from
that in which the vocabulary of 'true'and 'false'is used in discussing ordinary matters
of fact, in everyday life or in science or history, or even literary criticism. Using
Bernard Williams's terminology one might say that words such as 'true'and 'false'

end p.25

are not used 'substantially'when used like this.

'Consistency and Realism', Proceedings of the Aristotelian Society, suppl. vol. 40 (1966). Also 'The

Truth in Relativism'.

This is not, of course, to say that there is something wrong with the usage, but it does
raise doubts about the weight that can be placed on it in discussions of relativism.
The linguistic facts were appealed to in the attempt to show that moral judgements
could not be relativistic as some judgements of taste seem to be. Yet if we suppose,
as a kind of thought experiment, that the same linguistic possibilities exist in the case
of these other judgements we see that their relativism is left unchanged. In this new
situation it would be possible for an individual to reject as 'false' everything that other
members of his society said about good-looking faces, and such things, and it would
be possible for one society so to describe the views of another society however far
apart their judgements. The important point is that substantial truth would still belong
only where common standards were in some sense presupposed; it would still be

right to deny that there was any substantial truth belonging to the standards of any
particular community.
According to the argument just presented relativism is true in a given area if in that
area all substantial truth is truth relative to one or other of a set of possible standards.
And it is now possible to see that individualistic subjectivism may itself be a form of
moral relativism. (Perhaps we would call it a limiting case.) For even if the truth of
moral judgements is not relative to local community standards it (the substantial
truth) could still be relative to the standards of the individual. This is how it is, in
effect, in emotivist and prescriptivist theories, since these theories deny the presence
of objective criteria, or any objective method by which differences between
individuals with radically different basic moral principles could in principle be
resolved. If these theories are correct, anyone who queries the truth of a moral
judgement, and still possesses the resource of testing it by his more basic moral
principles, uses 'true' substantially; but beyond this point he does not. It follows that
the emotivist or prescriptivist is committed to a form of relativism, however little he
may like the label. Stevenson, who claimed to have refuted moral relativism, turns
out to be himself a kind of moral relativist.

end p.26

This account of relativism is gravely deficient in so far as it depends on the idea of
substantial truth, and gives only merest indication of what this is. Nevertheless there
is enough in what has been said about that, and in the comparison with judgements
of taste, to make it possible to enquire further into the implications of moral relativism;
and at this point I want to refer to a discussion of the topic appearing in Walter
Stace's book The Concept of Morals in 1937.

(London: Macmillan, 1937), chs. 1 and 2.

Stace takes the relativist to be one who denies that there is any single objective
standard of morals, and this is, of course, in line with what we have just been saying.
But he also attributes to him two other beliefs: firstly that the very same action that is
right in one country or at one period may be wrong in another, and secondly that if a
man thinks something right it is right for him.
Let us consider the first proposition, once more thinking about the analogy provided
by our relativistic judgements of taste. It was seen that such assertions may be true
by the standards in operation in one social context and false by those equally well
established in another; but it would be wrong to infer, for example, that some men
are good-looking in a certain place but not elsewhere, as if their complexions
suffered when they moved. And it would be similarly misleading to say that a man
might become less good-looking as time went on not because he got older but
because standards changed. Nor could two sets of standards be employed
simultaneously, to make the same man at the same time both good-looking and ugly.
This would be no better than declaring in mid-Atlantic, half-way between southern
Italy and the US, that a man of a certain height was both tall and short. If, for a
certain type of judgement, a local standard automatically comes into force, every
proposition of this form will presuppose one rather than another. Even with the
limiting case of relativism, where the reference is supposed to be only to the
standards of the speaker, it will be impossible to employ two sets of standards at
once.
Stace is wrong, then, in thinking that the moral relativist is committed to the first of
the two propositions listed above. But is he wrong also

end p.27

about the second? This is much more interesting, and more debatable, but we have
first to try to get clear about what it means to say that if a man thinks something right

it is right for him. It could, I suppose, be taken as denying any distinction between 'A
thinks it right to do X'and 'It is right for A to do X', rather as we might deny the
distinction between 'A thinks the temperature in the room pleasant'and 'The
temperature in the room is pleasant for A'. So interpreted the proposition would deny
the possibility of error about matters of right and wrong, and this would seem to make
it obviously false. For even on subjectivist theories a man may apply his own
standards wrongly, and hence there is a possibility of 'correction' that is not simply a
change of mind. Leaving these possibilities aside we must, however, give further
attention to Stace's second formulation of relativism, because it expresses a thought
that does much to account for the seduction of moral relativism. Surely, we may
think, an agent must be doing what he ought to do so long as he is following his
conscience? For how could he act rightly if he did not?
To see that this argument is specious we need to consider separately the
propositions (1) that someone acts badly in going against his conscience and (2) that
he acts well in following its dictates. It seems to be taken for granted by
contemporary moralists that the two propositions are on the same footing, and do not
need to be considered separately; but this is curious given that Aquinas, who would
strenuously have disagreed, wrote powerfully against this view.

Summa Theologica, 1a 2ae, q. 19 aa. 5-6 (First Part of the Second Part, question 19, articles 5-6).

He argues that anyone acts badly if he goes against his conscience, whatever it may
be that it tells him to do, so that even an erring conscience 'binds'. It does not follow,
however, that whoever follows his conscience necessarily acts well, and Aquinas
insists that he will also act badly when he does the things that are evil, even if he
thinks them good. Of a man such as Himmler, who seems to have believed that he
should

order the gassing of Jewish men, women, and children, Aquinas would have

said that that so long as he held that opinion he could not but act badly. For either he
would spare the lives of those in the concentration camps and so go against his
conscience or else he would have them killed and so do wrong. To the suggestion
that a

end p.28

parallel argument could be used to show that whichever Himmler did he would act
well, since either he would act well in following his conscience or else in doing what
was good, Aquinas would have replied that good and evil are not symmetrical as that
argument requires. Even if someone does what he should do in saving or sparing
lives, he nevertheless acts badly if he believes this a bad thing to do and so goes
against his conscience. And he also acts badly even in following his conscience if he
does something that is in fact bad. For a single defect is enough to make something
defective whereas a single merit does not suffice to make it good.

Ibid. 1a 2ae, q. 6 a. 8, and q. 19 a.6 (First Part of the Second Part, question 6, article 8, and question

19, article 6).

This surprising principle turns out to be one with which we must agree, and agree
quite generally. A house is a bad house if it is either badly designed or badly built; it
is a good house only if good in both ways. An apple is bad if it is either worm-ridden
or tasteless, and good only if free from either defect. This is not of course to say that
any small defect is enough to make us call something bad: the point is that there is
here substantial asymmetry of the kind that Aquinas's argument requires.
It is possible, therefore, to maintain with Aquinas that although the erring conscience
binds it does not excuse. One may consistently say that if a man thinks he is doing
wrong he is acting badly, and also that he may be acting badly even if he thinks that
he is doing what he ought to do. The case is parallel to that of any judge who has two

ways of giving an unjust judgement, firstly by giving one that (even if mistakenly) he
believes to be unjust, and secondly by giving one that really is. And similarly anyone
has two possible ways of demeaning himself: he can do what he sees as low or
unworthy or demeaning, or he can do something to which these terms do, though he
does not see it, apply. It follows that in discussing Stace's second formulation of the
relativist's position, and the claim to the truth of relativism as thus defined one does
indeed need to deal separately with the two propositions (1) and (2) above.
That someone who does what he thinks he should not do acts badly cannot, I think,
be denied. One might say that there could not be a more radical defect than that of
being prepared to do what one believes to be wrong. A person of whom this is true is
like an archer who does not even

end p.29

aim his arrow at the place where the target seems to him to be: it will be the merest
chance if he does what is good when he is doing what he sees as bad. Proposition
(1) should therefore be accepted. To say this is not, however, to take the smallest
step towards the doctrine of the relativist who is supposed to believe also in
proposition (2) in opposition to an anti-relativist who insists that there are objective
standards of morality independent of an agent's beliefs.
We should therefore return to the question of whether or not there are indeed
objective standards by which kinds of actions can be judged. Aquinas's discussion of
the erring conscience has demolished a certain specious argument in favour of moral
relativism, but has here taken for granted rather than established the kind of
objectivity across personal and communal boundaries that relativism denies. Might
we not continue with moral judgements as we now have them while admitting that
there are no objective criteria to which appeal can be made in case of radical cross-
cultural disagreement? After all we speak of Nureyev's good looks and claim that
blue and green are really harmonious colours without worrying about the fact that by
the standards of other times or places these assertions might be false, and yet their
standards no worse than our own. Truth by local standards is here the only truth that
we have, and it is this truth that we tacitly claim for our opinions when we express
them. If there is something that makes this kind of accommodation mistaken in the
case of moral opinions it must be brought to light.
Some will no doubt argue that moral judgements are different from what I have called
'judgements of taste' in so far as we must wish to see others adopt our own moral
system, whereas we do not mind what they think about good-looking faces, delicious
food, and other matter of taste: everyone must necessarily preach his own moral
opinions, refusing to live and let live, and he therefore cannot say that opposing
views are as good as his own. That this is an invalid argument can be seen, I think,
by considering the case of imperatives. If two persons issue opposing orders each
will, in general, want his own instruction obeyed, but there is here no built-in standard
by which one is to be judged 'true'and the other 'false'.

end p.30

The question is, therefore, whether we have the same reason to accept relativism
with regard to cultures with very different codes of right and wrong as we have to
accept it where there is such divergence in matters of taste. This it seems to me that
we do not. For our starting point there was the thought that at least some very
general judgements of taste could be identified through any amount of variation in the
application of the key concepts through the relevant domain. I myself have frequently
argued that by contrast such variation cannot be postulated in the case of moral
judgements, because the thought of moral goodness and badness cannot be held

steady through any and every change in the codes of behaviour taught and in their
grounds.

See e.g. P. R. Foot, 'Moral Arguments', Mind, 67 (

1958

), and 'Moral Beliefs', Proceedings of the

Aristotelian Society

, 59 (

1958-9

From this it follows that not everything that anyone might want to call 'a moral code'
should properly be so described. And this shows incidentally that hypotheses about
de facto cultural relativism are not totally independent of moral theory. Even if an
anthropologist is inclined to call a certain code a moral code, and to go on to talk
about a morality radically different from our own, it does not follow that we should
accept this way of describing the phenomena. An anthropologist may be as confused
or prejudiced as anyone else in applying words such as 'morality' to the teachings of
an alien culture.
I shall assume that even general moral terms such as 'right' or 'ought' are restricted,
to a certain degree, in their extension, at least at the level of basic principles. It is not
possible that there should be two moral codes the mirror images of each other, so
that what was considered fundamentally right in one community would be considered
wrong at the same level in the other. It seems that some considerations simply are,
and some are not, evidence for particular moral assertions. Nevertheless it does not
look as if a correct account of what it is to have a moral thought, or a moral attitude,
or to teach a moral code, will suffice to dismiss relativism throughout the moral
sphere. Even if some moral judgements are perfectly objective, there may be others
whose truth or falsity is not easily decidable by criteria internal to the subject of
morality. We may suppose,

end p.31

I think, that it is clearly an objective moral fact that the Nazi treatment of the Jews are
morally indefensible, given the facts and their knowledge of the facts. The
Nazis'moral opinions had to be held on grounds either false or irrelevant or both, as
on considerations about Germany's 'historic mission', or on the thought that genocide
could be a necessary form of self-defence. It was impossible, logically speaking, for
them correctly to argue that the killing of millions of innocent people did not need any
moral justification, or that the extension of the German Reich was in itself a morally
desirable end. Yet after such things have been said the problem of moral relativism is
still with us.

Compare P. R. Foot, 'Morality and Art', ch.

in this volume.

Even if the fact that it is morality that is in question gives us some guaranteed
starting-points for arguments about moral right or wrong, how much is this going to
settle? Are there not some moral matters on which, even within our own society,
disagreement may be irreducible? And is it not possible that some alien moral
systems cannot be faulted by us on any objective principles, while our moral beliefs
can also not be faulted by theirs? May there not be places where societies simply
confront each other, with no rational method for settling their differences?
In the most penetrating critique of relativism that has appeared for many years
Bernard Williams has recently argued that this is so, and he bases his belief in moral
relativism on this possibility. He thinks that although the vocabulary of appraisal can
be used 'substantially' in the consideration of the moral beliefs of some alien cultures,
this is not always the case. Is he right about this? He supposes that 'substantial'
employment of terms such as 'true' or 'false' can occur only where a system is in 'real
confrontation'with our own; and real confrontation turns out to be given by the fact
that we could, without suffering hallucination or otherwise losing our grip on reality
and rationality, come to hold those beliefs. What is puzzling is why he thinks this

relationship to be either necessary or sufficient to give substance (he says, perhaps
significantly, 'point or substance') to the vocabulary of appraisal.

'The Truth in Relativism', sect. 2.

If we think back to the judgements of taste discussed earlier in this lecture we see
that anything of this kind would be irrelevant there. By some process of

end p.32

acculturation we could, no doubt, come to hold quite different opinions about good-
looking faces, delicious foods, and combinations of colours. But this does not mean
that we can criticize the systems which are a real option for us.'True'and 'false'get a
substantial use where there are objective criteria, or at least methods of some kind
for settling disputes, and this seems to have nothing to do with 'real confrontation' as
defined by Williams.
Nevertheless, Williams is surely right in thinking that if some societies with divergent
moral systems merely confront each other, having no use for the assertion that their
own systems are true and the others false except to mark the system to which they
adhere, then relativism is a true theory of morality. Yet at this point one may become
uneasy about his reasons for saying that relativism is true. For it seems strange to
suggest that there is any society whose values we can identify without being able to
set them critically beside our own, and our own beside theirs. Some parts of the
moral vocabulary do indeed seem unusable when we are considering very alien and
distant communities. For instance it would be odd then to talk in terms of the
permissible and the impermissible, simply because language of this kind cannot
venture very far from actual sets of permissions and prohibitions. But this does not
mean that we cannot in any way judge the moral rules and values of societies very
different in this respect from our own. Granted that it is wrong to assume identity of
aim between peoples of different cultures; nevertheless, there is a great deal that all
men have in common. All need affection, the cooperation of others, a place in a
community, and help in trouble. It isn't true to suppose that human beings can
flourish without these things—being isolated, despised, or embattled, or without
courage or hope. We are not, therefore, simply expressing values that we happen to
have if we think of some moral systems as good moral systems and others as bad.
Communities as well as individuals can live wisely or unwisely, and this is largely the
result of their values and the codes of behaviour that they teach. Looking at these
societies, and critically also at our own, we surely have some idea of how such things
work out and why they work out as they do. We do not have to suppose it is just as
good to promote pride of place and the desire to get an advantage over other men as
it is to have an

end p.33

ideal of affection and respect. These things have different harvests, and
unmistakably different connections with human good.
No doubt it will be argued that even if all this is true it leaves moral relativism
substantially intact, since objective evaluation of moral systems can go only a little
way, and will come to an end before all the radical disagreements are resolved. One
wonders, however, why people who say this kind of thing are so sure that they know
where discussions will lead and therefore where they will end. It is, I think, a fault on
the part of relativists, and subjectivists generally, that they are ready to make
pronouncements about the later part of moral arguments, and moral reviews, without
being able to trace the intermediate steps. Nor is it that they just do not bother to take
the whole journey; for there are reasons why they are not able to. One of these has
to do with conventions about moral philosophy, conventions that forbid the
philosopher to fill chapters with descriptive material about human nature and human

life. It isn't supposed to be part of his work to think in the somewhat discursive way
that is suitable to reflections about the human heart, and the life of men in society. It
is, of course, a kind of decency that keeps moral philosophers in the analytic tradition
away from the pseudo-profundity that is found in some philosophies as well as in
vulgar preachers. Yet it may be that they have to do this work, and do it properly,
before they will know the truth about divergent moralities and values. And we have,
after all, a rich tradition of history and literature on which to draw.
There is, however, another reason why moral philosophers tend to give only a sketch
of the beginning and end of discussions of the values of different societies; and this
has to do with a gap in our philosophical understanding. Perhaps it would be better to
speak of a series of gaps, of which I shall give instances in the next few paragraphs.
My thought is that there are some concepts that we do not understand well, and
cannot employ competently in an argument, but which are, unfortunately, essential to
genuine discussions of the merits of different moral systems.
Let me give some examples of the kind of thing that I have in mind. I would suppose,
for instance, that in some fundamental moral enquiries we might find ourselves
appealing to the fact that human life is of value. But do we really understand this
thought? Do we know what we mean by

end p.34

saying that anything has value, or even that we value it, as opposed to wanting it or
being prepared to go to trouble to get it? I do not know of any philosopher living or
dead who has been able to explain this idea. And then again we are likely to find
ourselves talking about happiness, which is a most intractable concept. To realize
that one does not understand it one has only to try to explain why, for instance, the
contented life of someone on whom a prefrontal lobotomy has been performed is not
the happy life, or why we would count someone as unfortunate rather than endowed
with happiness if he were tricked into thinking he was successfully spending his life
on important work when he was really just messing around. That we do not
understand the concept of happiness is shown, once more, by the fact that we are
inclined to think stupid thoughts about the idea of great happiness, as if it were
simply extreme and prolonged euphoria. That great happiness depends on its objects
is a surprising idea once we understand this, as it should be understood, as a
conceptual not a causal matter. It seems that great happiness, unlike euphoria or
even great pleasure, must come from something related to what is deep in human
nature, and fundamental in human life, such as affection for children and friends, the
desire to work, and love of freedom and truth. But what do we mean by calling some
things in human nature deep, and some things in life fundamental? In one way we
know this, because we are able, for instance, to understand a man who says at the
end of his life that he has wasted his time on 'things that don't matter'. But what are
things that 'matter' if they are not the trivial things on which we spend so much time?
Clearly such questions are relevant to fundamental discussions of the moralities of
other societies and our own. It is impossible to judge a society's morality if we cannot
talk about its values, and we must be able to handle the thought of false values if we
are to say what is wrong with a materialistic society such as ours. But what is it to
have false values if it is not to think too highly of things that do not matter very much?
It seems, then, that we are all at sea with some of the ideas that we are bound to
employ in any real discussion of divergent moralities. What we tend to do is to ignore
these ideas, and pretend that the debate can be carried on in other terms. But why
should this be supposed possible? With

end p.35

other, more jejune, concepts we shall get a different discussion, and from it we
cannot draw conclusions about how the first would end. Moving from one to the other
we are merely guessing at results, and this is, I think, exactly what happens in many
arguments about moral relativism. Personally I feel uncomfortable in these
arguments, and perhaps this is because I am advancing opinions on the basis of a
guess. The practical conclusion may be that we should not at the moment try to say
how far moral relativism is true or false, but should start the work farther back.

end p.36

Moral Realism and Moral

Dilemma

Philippa Foot

Two articles written by Bernard Williams in the middle sixties have recently been
receiving attention in the journals, and as these articles argue against moral realism
the idea is abroad that moral realism is under attack. Moreover, the basis of the
attack seems to be new. Formerly, emotivists and prescriptivists started out from
pervasive features of moral language to draw a contrast between moral judgements
and descriptions of the world; but the latter-day anti-realism bases itself rather on a
set of special cases: those in which moral judgements seem to collide. The
suggestion is that moral realism or cognitivism (which are not here distinguished)
cannot do justice to the facts of moral dilemma or conflict. So in 'Ethical
Consistency'

Proceedings of the Aristotelian Society, suppl. vol. 39 (1965).

Williams insisted that the feelings we have in situations of moral conflict show that
the 'structure'of moral judgements is unlike that of assertions expressing beliefs. In
'Consistency and Realism'

Proceedings of the Aristotelian Society, suppl. vol. 40 (1966). Both articles are reprinted in Bernard

Williams, Problems of the Self (Cambridge: University Press, 1973), and subsequent references are to
this volume.

he argued that we can tolerate inconsistency in moral

end p.37

principles though not in assertions, and that this is explained by the fact that it is the
concern of the latter but not of the former to reflect an 'independent order of things'.
I shall suggest that in each of his papers Williams's main line of argument is
mistaken. Much of what he says in 'Ethical Consistency' about moral conflicts not
being resolvable 'without remainder'seems to me to be true, but in no way inimical to
moral realism. Much of what he says in 'Consistency and Realism' about the
tolerability of conflicting moral principles also seems to be true, but again the anti-
realist or anti-cognitivist inference seems mistaken. This is what I shall try to show.

To open the discussion we must, of course, ask what it is that Williams, and others
who have recently written about moral conflict or moral dilemma, are talking about.
We notice at once that cases of 'moral dilemma'are not necessarily ones in which
anyone is in a dilemma about what to do; and that the conflict is not the war that goes
on in someone's mind when he is torn between alternatives. For while in many
favourite examples the protagonist is torn; like Agamemnon who must sacrifice his
daughter to save his campaign, or Sartre's youth caught between his duty to his
mother and to the cause of freedom, it is unlikely that anyone who has to break a
promise to see a friend, in order that he may save a life, should be in conflict about
what to do. Yet the latter is also treated in this discussion as an example of moral

conflict. The conflict in question is therefore between principles such as keep
promises

and save lives, not the conflict that may be produced in someone's mind by

these clashes of principle.
In situations of moral conflict as thus understood one principle enjoins one action and
another another, and it is impossible that the agent should do both. Usually the
impossibility is fairly and squarely an empirical impossibility, and it is often up to the
agent to rack his brains for a way out before declaring that the conflict is real. But
sometimes the mere descriptions of the obligations are such as to rule out joint
compliance, as when one has promised to say nothing and also to tell all. In one

end p.38

way or another joint compliance is ruled out, and this is what matters in the present
context. Williams prefers to describe conflict situations in terms of a and b and the
impossibility of doing both a and b. For the sake of convenience I shall, however,
normally talk about the fact that someone ought to do a and ought to do not a. I shall
treat "X ought not to do a"as a mere stylistic variation on "X ought to do not a". With
"It is not the case that X ought to do a", which is of course quite different, I shall not
be directly concerned.
As already mentioned Williams argues in 'Ethical Consistency' that cognitivist
accounts of moral judgement cannot do justice to the facts of moral conflict,
especially as these facts concern the feelings attending choice in conflict situations.
He writes:
It seems to me a fundamental criticism of many ethical theories that their accounts of
moral conflict and its resolution do not do justice to the facts of regret and related
considerations: basically because they eliminate from the scene the ought that is not
acted upon. A structure appropriate to conflicts of belief is projected on to the moral
case; one by which the conflict is basically adventitious, and a resolution of it
disembarrasses one of a mistaken view which for a while confused the situation.
Such an approach must be inherent in purely cognitive accounts of the matter; since
it is just a question of which of two conflicting ought statements is true, and they
cannot both be true, to decide correctly for one of them must be to be rid of error with
respect to the other—an occasion, if for any feelings, then for such feelings as relief
(at escaping mistake), self-congratulation (for having got the right answer), or
possibly self-criticism (for having so nearly been misled).

Problems of the Self, 175-6.

Williams says that moral conflicts are not all soluble without remainder and that this is
implied by the regret that we sometimes feel in conflict situations even when we are
convinced that we acted for the best. Two principles compete with each other for the
guidance of our conduct and the one whose claim is rejected is not simply eliminated
like a belief that we decide is false. The reality of the claim that we judge less
pressing goes on making itself felt in the form of regret, and in the form of other
phenomena such as our willingness to make it up to anyone who is

end p.39

injured by our choice. In this, he thinks, moral conflicts are more like conflicts of
desire than of belief. If I decide that one of two conflicting beliefs is true the other
cannot substantially survive the decision, because to decide that a belief is untrue is
to abandon that belief. A rejected desire, on the other hand, may survive the decision
not to satisfy it, and even when opportunity for satisfaction is past it may reappear in
the form of a regret for what was missed.
Thus Williams believes that one cannot accept a picture of moral judgement which as
he puts it 'makes it a necessary consequence of conflict that one ought must be

totally rejected in the sense that one becomes convinced that it did not actually
apply'.

Problems of the Self, 184.

And this account he had said, in the passage quoted earlier, to be a necessary
consequence of cognitivist theories of ethics. I shall first consider the argument from
feelings to 'remainder', then the facts about what is left over after the solution of a
moral conflict, and finally the implications, if there are any implications, for the debate
about moral realism.
The feelings that Williams thinks so significant are the feelings of regret or
'distress'that someone may feel even when he is convinced that his choice of actions
was morally justified—that he was 'acting for the best'. The regret or distress is
supposed to show that the agent thinks he was doing something he ought not to
have done, e.g. in breaking a promise, even where he also thinks that in the
circumstances it was right to break the promise. Opting in favour of the second ought
leaves the other intact. Against those who would call the feelings irrational Williams
says that they do the agent credit by showing that he takes promises seriously. He
also insists that the regrets spring from the agent's thought that he has done
something that he ought not to have done and not from some other distressing
feature of the situation.
The form of this argument is surely strange. It is not, of course, to be denied that
feelings such as regret are 'propositional', so that to feel regret is at least to feel as if
something in some way bad has happened. But it does not follow that it has
happened, and perhaps not even that the subject

end p.40

thinks it has, since one may say, e.g., 'I can't get away from a feeling of regret,
though I know I haven't actually lost anything at all'. It is impossible to move from the
existence of the feeling to the truth of the proposition conceptually connected with it,
or even to the subject's acceptance of the proposition. Nor will it help to argue, as
Williams does, that if we call such feelings irrational we must be using the description
non-pejoratively. He thinks it creditable for someone to feel 'moral distress' when
making a choice in a conflict situation; to which one would reply that this is neither
here nor there. There are plenty of feelings which are irrational without being
discreditable, as for instance feelings of guilt about giving away the possessions of
someone lately dead. It would obviously be wrong to conclude from the fact—the
normality almost—of such feelings that there was indeed some element of
wrongdoing involved.
It is a mistake then to think that the existence of feelings of regret could show
anything about a remainder in cases of moral conflict. The feelings are rational
feelings only if it is reasonable to think that given a conflict situation there is
something regrettable or distressing even in a choice that is clearly right. What we
find is, I think, that there may indeed be a 'remainder' in the shape of obligations
unfulfilled, and things left undone which it is correct to say that we ought to have
done. But whether it is always regrettable or distressing when obligations are
unfulfilled or things left undone which ought to be done is more doubtful. I shall come
back to this point after discussing the obligations and oughts which are left over when
a moral conflict is resolved.
Suppose a case in which I have promised to do an action a and promised to do an
action b; say to be best man at A's wedding and also at B's. But bad luck A and B fix
their weddings for the same day and I can't attend both. So one of my promises must
be broken, and as we are considering resolvable conflicts we may suppose that for

some reason my promise to A has clear precedence over my promise to B.
Nevertheless, I promised B, and nothing has happened to release me from this
promise. I have an obligation to him which, when I opt for A's wedding, I decide not to
fulfil. I still have the obligation and it doesn't much matter at what

end p.41

point it will be right to say that I had it rather than that I have it. In one form or another
the obligation stands, unless B releases me from it before the time for fulfilment is
past.
There is, therefore, the possibility of saying truly that I have an obligation to do two
things that cannot both be done. I have an obligation to do each of them, although I
do not of course have an obligation to do them both.

See Problems of the Self, 181-3 for Williams's denial of what he calls 'the agglomeration principle'.

It can similarly be true that I ought to do them, though it is less common to speak of
two things each of which I ought to do but both of which I cannot do, than of two
mutually exclusive obligations. However, the possibility of employing "ought"like this
reminds us that the area of conflict-without-inconsistency extends beyond that part of
morality which has to do with what are strictly obligations. Moreover, we find the
phenomenon outside morality, since it makes perfectly good sense to say, when
pressing business has given one overriding reason to go to town, that one
nevertheless ought to be at home nursing one's awful cold. Indeed it may seem
surprising that anyone should ever have denied that I can have an obligation to do a
and an obligation not to do a, or that I ought to do a and ought not to do it. Why is this
harder to accept than the fact that I can have two engagements that conflict?
"Engagement" like "obligation" and "ought" is an 'action-guiding' term: if people did
not look in their books or consult their memories and say things like 'I have an
engagement so I must go . . . ' the concept of an engagement would not be known.
But I can have conflicting engagements as I can make conflicting promises. Given
the possibility of lying promises and of memory lapses I may even have promised
that I will do a and that I will not. If I cannot have an engagement to do a and not to
do a this is merely because (i) we do not have negative engagements and (ii) in
"engagement to do a"the context of reference of "a"is arguably opaque. I can only too
easily have an engagement to do a and an engagement to do b, where I can't do
both. Propositions about engagements are what I shall call "type 1" propositions. We
shall see later on that "obligation" and "ought" can appear also in another type of
statement, but for the moment the comparison with promises and engagements is to
be kept in mind.

end p.42

Why is there resistance to the idea that an obligation or ought may be overridden
without being destroyed? I think that there are two reasons.
The first reason is that a certain kind of example sticks in one's gullet. Suppose, for
instance, that some person has an obligation to support a dependent relative: an
aged parent perhaps. Then it may be that he ought to take a job to get some money.
The obligation produces an ought related to means. So far so good. But what if the
only means of getting money is by killing someone? The obligation to refrain from
taking someone's life—to refrain from interfering with him in this way—is stronger
than the obligation to give aid to the parent, and it is therefore clear what should be
done. Should we nevertheless say that although the agent ought not to kill he also
ought to kill, since oughts which are put out of action by stronger oughts are not
thereby destroyed? Surely this ought is destroyed by the superior injunction against
taking life?

We need an explanation of why it is not the case that the son or daughter ought to kill
to get the money, but an explanation which will not interfere with the general principle
that oughts that are inactivated may nevertheless stand. We find such an explanation
in the fact that in the problem case the ought would have been attached to the killing
only because killing was a means to the fulfilling of an obligation. For the only things
that count as means are possible actions. If some obligation of mine could be fulfilled
only by my flying up to the ceiling we do not therefore say that I ought to fly up to the
ceiling. This may seem irrelevant, for we think 'After all it is possible for the agent to
kill and get the money'. But then we have failed to notice that it is not only natural law
possibility that matters but also moral possibility. When we say 'I can't do such and
such' we do not necessarily mean that there is nothing in our power that would bring
it about. Often we have no idea whether there are some steps we could take if only
no holds were barred. In our example killing is not a possibility and so there is no
question of treating it as a means which ought to be taken to fulfil the obligation to
support the aged parent. The obligation stands though we cannot discharge it; but it
does not make the killing into something that ought to be done. Puzzling as moral
modalities are, we have to take them into account if we are going

end p.43

to understand the way concepts such as obligation and ought work in cases of moral
conflict.

The relation between obligation and moral possibility is complex. Where an obligation is overridden

we may say that it—the obligation—could not be fulfilled. ('I was unable to fulfil my obligation.') But
while original obligations may stand it seems that derivative obligations are aborted.

The second reason why it is hard to see that conflicting ought statements ("ought
a

"and "ought a") are consistent is that there is another kind of ought statement for

which this is not true. When "ought a" and "ought  a"are both of the second kind, are
both as I shall say used in type 2 ought statements, they are not consistent. This
distinction between types of statements is essential to my argument.
What is a type 2 ought statement? What is it that makes "ought a"(2) inconsistent
with "ought  a"(2), although "ought a"(1) is consistent with "ought  a"(1)?
The explanation is that type 2 ought statements tell us the right thing to do, and that
this means the thing that is best morally speaking, or speaking from whatever other
point of view may be in question. It is implied that for one for whom moral
considerations are reasons to act there are better moral reasons for doing this action
than for doing any other. As this cannot be true both of a and of  a, "ought a"(2) is
inconsistent with "ought  a"(2). "Ought a"(2) is not, however, inconsistent with "ought

"(1). I can have reason not to do something and yet have better reason to do it

than I have to do anything else.
From the fact that "ought a"(1) is consistent with "ought  a"(1) but that "ought a"(2) is
not consistent with "ought  a"(2) it follows that if we are to keep that intuitively sound
idea that what one ought to do one is permitted to do it is in terms of type 2 ought
statements and not those of type 1 that permissibility must be defined. So "a is
permissible" = def. "It is not the case that a ought not to be done" (type 2 ought). And
"ought a"(2) does imply  (ought  a)(2).
The division into type 1 and type 2 propositions, as I am describing it, belongs also to
some other action-guiding statements. The test of whether an action-guiding
predicate appears, or is here appearing, in statements of one type or the other is that
it does or does not make sense

end p.44

to add φ( a) to φa. By this test we would determine that "dangerous"on its own
appears only in type 1 statements, since the assertion that it is dangerous to do a is

always consistent with the assertion that it is dangerous not to do a. (It may or may
not be more dangerous to do a.) If it is dangerous to pick up a poisonous snake it
may nevertheless be dangerous not to pick it up if that is the only way of getting rid of
it. By contrast "imprudent" is a predicate appearing only in type 2 propositions: if it is
imprudent to do a it cannot be imprudent not to do it.
From this last example we might conclude that a type 2 statement is one asserted on
an "all things considered"basis, while a type 1 statement is not. But this would be
wrong. "All things considered it is dangerous" contrasts with "Prima facie it is
dangerous"and both are about a dangerousness of doing something which is
compatible with the dangerousness of not doing it. And type 1 obligation statements
can also be said either to be prima facie true or to be true all things considered. For
instance there is a prima facie case for thinking that I have an obligation to support
someone in old age given that he is my father. But perhaps he deserted my mother
before I was born, and then I may say that all things considered I have no obligation
to him. But even if, in rather different circumstances, I think that all things considered
I do have an obligation to this man it does not follow that the right thing to do would
be to support him; because my obligation to my children may override my obligation
to him. So it was only a type 1 obligation statement that was said to be, all things
considered, true.

In his article 'Prima Facie Obligations', in Z. van Straaten (ed.), Philosophical Subjects: Essays

Presented to P. F. Strawson

(Oxford: Clarendon Press, 1980) John Searle seems to miss the difficulty

of identifying what I call type 2 propositions as 'all things considered' statements.

Yet we do naturally indicate that we are employing "ought" in a type 2 statement by
talking about what ought all things considered to be done. Why is this? I think it is
simply because the same word "ought" is used in both kinds of statements, and type
2 propositions take as evidence all the available type 1 statements about the same
subject together with any principles for deciding priorities. Therefore, in moving from
the consideration of any type 1 ought proposition to a type 2 ought proposition about
the same

end p.45

subject we can mark the transition by asking 'And what all things considered ought
we to do?'
We now have a clue to the relationship between type 1 and type 2 statements about
what ought to be done. A type 1 proposition says what is the case not what prima
facie

is the case. But a reference to what it states to be the case can appear in a

prima facie

statement about a type 2 predication. So the fact that something is

dangerous shows that prima facie it is imprudent to do it. And the fact that I have an
obligation (1) to do something shows that prima facie I ought (2) to do it.
We have now seen one definite way in which even the clearest solution of a moral
conflict, expressed in a type 2 ought statement which nobody doubts, nevertheless
can leave 'remainder'. It may be the case that a 'conflicting' type 1 proposition about
an obligation, or about what the agent ought to do, or to have done, is true. Let us
now return, briefly, to the question of feelings such as regret or 'distress'. Does the
truth of a type 1 obligation or ought statement make such feelings rational? This, as I
said before, does not seem to me as clear as Williams and others suppose.
Take, for instance, the breaking of a promise. One has promised to meet someone
but must instead take an accident victim to hospital. Obviously one regrets it if there
isn't time to let the promisee know and he is incommoded. But this is regret for a
consequence, not regret for the breaking of a promise as such. So, to isolate the
latter we shall suppose that things turn out splendidly all around; the promisee does
not have a moment's annoyance, and meets his future beloved, or someone who

offers him a job, while standing at my door. Are we to say that nevertheless in the
general rejoicing there should be an element of distress (moral distress) because
after all a promise was broken and that is something bad, and therefore regrettable?
To this suggestion one hardly knows what to reply. A charitable man would wish for a
world in which promises did not have to be broken (as he would wish for a world in
which there were no earthquakes) but this is because the breaking of promises, even
when necessary, usually has bad effects. And we may perhaps suppose that a moral
man must regret the wickedness of the world that contains wantonly broken promises
and must regret it even when nothing bad follows. But our case is one in which
neither bad consequences nor

end p.46

wickedness is there to be regretted. Someone who nevertheless felt distress would
seem to many of us rather foolish, though one can of course imagine a society in
which someone was held to incur shame or be 'tainted' by having to break a promise
even for the best of moral reasons. And we ourselves might feel a distress that we
thought rational if for instance we had had to reveal a secret that we had promised
we would never tell, even if no harm had come of it or could come of it. If the secret
were something that its possessor had wanted very much to guard then even if his
reputation had been enhanced rather than damaged by the telling of it, and even if he
were now dead, one might still feel that it was something regrettable that one had to
do, and one might hate to do it. The area seems to be one of uncertainty, but
perhaps this very fact makes the argument from feelings to 'remainder' in the solution
of moral conflicts a bad one. And it is in any case unnecessary since the language
clearly allows us to use "obligation" and "ought" as we use "engagement"in type 1
statements.
Against one argument from feelings one must definitely protest. This is an argument,
not given by Williams himself, about feelings of guilt. It may seem obvious that no
one can be guilty if he acts 'for the best', but this has recently been denied by Bas
van Fraassen and by Ruth Marcus, though on slightly different grounds. Van
Fraassen questions the connection between guilt and what is in the agent's power,
suggesting that it is denied in the doctrine of original sin.

Bas C. van Fraassen, 'Values and the Heart's Command', Journal of Philosophy, 70/1 (Jan. 1973).

Ruth Marcus takes a line which seems to me more interesting.

Ruth Barcan Marcus, 'Moral Dilemmas and Consistency', Journal of Philosophy, 77/3 (Mar. 1980).

She says that while no one is guilty except by virtue of a choice he has made, this
does not rule out the possibility that he is in such a situation that he will be guilty if he
does some action and guilty if he does not, since he is free to choose to do it and
free to choose not to do it.
These arguments seem to me wrong, but they throw upon their critics the burden of
showing why guilt is not like this. After all shame is not, since no one has to be in any
way responsible for what brings shame on

end p.47

him, which might in Dostoevskian fashion be his craven or dissolute father. And then
there is the idea of 'dirty hands', which is perhaps nearer to what Marcus wants, since
one's hands get dirty only by one's voluntary actions—or so one may suppose—but
nevertheless the situation may be such that no one can emerge with clean hands
whatever he does. Perhaps he must either betray his friend's confidence or let an
innocent man be condemned through his silence. Either action seems shabby and
what Williams has called the moral 'disagreeableness' will not go away, even if there
is a clear solution and the agent is guided by it.

Bernard Williams, 'Politics and Moral Character', in Stuart Hampshire (ed.), Public and Private

Morality

(Cambridge: University Press, 1978), 62-4.

What is it then about guilt that makes this different? Since the argument is not about
a word, I don't want to dispute about the exact boundaries of the concept of guilt. It
will be enough to establish that there is some notion of fault which is such that no one
can be at fault both if he does a and if he doesn't do a, unless of course the fault is
imputed to him on account of some prior choice, or other past or present moral
failing.

A common case would be that in which he had carelessly, or with indifference, assumed two

obligations likely to conflict. I have throughout ignored this special case.

That there is such a notion of fault is shown by the fact that there is an imputation
against which not only physical or mental but also moral necessity is a shield. How
can this be denied? There is a clear place for the plea "I couldn't help it" (couldn't
help breaking the promise because I had to attend to the accident victim, and so on).
Nor does this plea simply plead mitigation, as if the offence of breaking the promise
was merely lessened. If you suffer because I cannot get to the appointment I have
with you, I say that I am sorry, meaning that I regret it; but if it was not my fault I do
not apologize, and I certainly do not have to 'make restitution' as some have
suggested. If I can't keep the appointment it isn't my fault that you suffer, and it
doesn't make any difference whether the necessity of breaking the promise was
physical, mental, or moral.
I do not know quite what Marcus, for instance, would say here. Would she deny the
whole procedure of refusing to admit moral fault on grounds of moral necessity? Or
would she say that it sometimes won't be

end p.48

allowed

, as if the plea will only do for cases in which no grave moral issues are

involved? It is hard to see why this would be more plausible than so restricting the
plea of physical or psychological necessity. The plea of moral necessity is that one
had to do a because it would have been worse not to. Since we are still dealing with
cases in which moral conflicts are resolvable, i.e. where it can be said that a is worse
than a or vice versa, and since this can be said no less in grave situations than in
others, either a is worse than a or the other way around. And there seems no basis
for driving a wedge between this and moral necessity, or between moral necessity
and the absence of fault. This is not of course to deny the suggestion that there are
some actions of which it cannot ever be said that I had to do them, pleading moral
necessity. But this is only to say that some things never can be such that it is morally
worse not to do them than to do them and my argument is neutral about this
possibility. What I am arguing is that if a is worse than a, then there is a kind of fault
that cannot be imputed to an agent who does a. Nor does it seem reasonable to deny
this kind of faultiness the name of guilt.
Let me now sum up the results of this first part of the discussion. We asked ourselves
a question about the solution of moral conflicts, namely whether they could be solved
only 'with remainder'. What emerged most clearly were some facts about obligations
and oughts: that an obligation is not annulled by being overridden, and that it is
possible to say that a subject ought to do something, even when a more pressing
claim makes it impossible for him to do that thing. I tried to sort out various
consistencies and inconsistencies which depended on the type of ought proposition
intended in the words "X ought to do a". It emerged that "I ought to do a"is in many
cases, though not always, consistent with "I ought not to do a".
How is all this relevant to theories of ethics, to cognitivism and non-cognitivist or
realism and anti-realism? How do things stand with Williams's arguments from the

facts about moral conflict to non-cognitivist? The first point to notice is a crucial
difference of terminology between Williams's articles and the foregoing discussion.
On the whole he talks about moral conflict, which is how he describes a clash
between moral principles where the clash is brought about simply by the

end p.49

facts. But he also talks about inconsistency in cases where the clash is between
logically incompatible alternatives, either a and not a or a and b, where the
descriptions of a and b are such that in no possible world are a and b both
performed. (He has a parallel vocabulary for describing desires which cannot both be
satisfied: one has either conflicting desires or inconsistent desires.) I find this piece of
terminology unhelpful for reasons that will become obvious as I go on. In my
vocabulary propositions, moral or otherwise, are inconsistent only if being either
contradictories or contraries they 'contradict'each other.
In the first of his two articles, 'Ethical Consistency', Williams's procedure is as follows.
He first contrasts conflicts of desire with conflicts of belief and then argues that moral
conflicts share characteristics with the former rather than the latter, which is what he
thinks his cognitivist opponent cannot admit. Moral conflicts are like conflicts of desire
in that they are not settled 'without remainder'. The decision to satisfy the first of two
desires when both cannot be satisfied does not in itself extinguish the second, which
may linger in the form of a desire, or of a regret for what was sacrificed. And similarly,
the decision to let one 'ought' or 'obligation' proposition guide one's conduct may
leave a remainder whose presence is, he says, indicated by moral regret or distress.
In the case of beliefs, however, the decision that one of two conflicting beliefs is true
means that the other has been abandoned. So, he thinks, the 'structure'of moral
judgements is unlike that of the assertions by which beliefs are expressed. And so
moral cognitivism is false.
What shall we say about this argument? First of all, that the structure of moral
judgements about what ought to be done is like that of statements about what is
desirable, and like expressions of desire, in so far as all allow of the "because of this
φx but because of that φ( x)" form of proposition, and therefore of "φx and φ( x)". But
this has nothing to do with cognitivism and non-cognitivism. If statements about what
is desirable or expressions of desire cannot be understood as being 'about the world'
this is not why they cannot be so interpreted. It was observed earlier that we may
have an engagement to do a and an engagement to do b where it is not possible to
do both a and b, and that the decision, however well justified, to keep one rather than
the other does not 'destroy'the other, which stands as 'remainder'. We should not,
however, be impressed

end p.50

by an argument against a cognitive theory of engagements based on the dissimilarity
in the matter of remainders between 'conflicts of belief'and 'conflicts of engagements'.
It is usually a plain matter of provable fact that I do or do not have an engagement; if
I say that I do have one my assertion is fully licensed by the evidence and however it
may be with moral judgements, statements about engagements undoubtedly express
beliefs. They differ from certain other cognitive affirmations as judgements about
what is dangerous differ from judgements about what is imprudent. They are action-
guiding propositions of a type 1 variety. But this, as the case of engagements shows,
is a difference found within the class of assertions (assertions about the world) and
has nothing to do with the demarcation of that class.
Williams thought that the cognitivist must believe that when two ought statements
conflict one is necessarily false, and would therefore find himself unable to explain
the fact of regret for an unfulfilled obligation such as a promise not kept. It does not

seem to have occurred to Williams that his opponent could simply allow the truth of"I
ought to do a"and "I ought to do a"and other consistent propositions whose
consistency is easily explicable on a 'because of this . . . but because of that . . . '
basis. And the reason it did not occur to him was, it seems, that he thought the
cognitivist was committed to a comparison between moral conflicts and conflicts of
belief. But why should the cognitivist ever accept such a comparison?

Samuel Guttenplan's otherwise strong article 'Moral Realism and Moral Dilemma', Proceedings of

the Aristotelian Society

, 80 (1979-80), is weakened, I think, by his failure to challenge Williams here.

Beliefs that conflict are beliefs that contradict each other, either directly or in the
context of other beliefs. But the whole point about statements about what is desirable
and what ought to be done is (for both) that there is a class of statements which
conflict

, in that they give conflicting guidance for action, but which nevertheless can

both be true

. The strange thing about what Williams wrote in 'Ethical Consistency'is

that a great deal of it seems designed to show exactly this: that moral conflict does
not imply 'contradiction'. It is as if he himself showed the cognitivist how to avoid the
very error he thinks the cognitivist must make.
I conclude that it is not 'inherent in purely cognitivist accounts' of

end p.51

moral judgement to think that only one of the conflicting ought statements can be true
in a case of moral conflict. The cognitivist can do justice to facts about 'remainders'
and about 'moral regret' or 'moral distress'exactly as well as anyone else.

So far we have been considering examples of moral conflicts which are resolvable,
though not 'without remainder'. In such situations there is an answer to the type 2
question "What, all things considered, ought N.M. to do?" although this answer may
leave standing any number of type 1 ought statements, some of which may conflict
with the first answer in that both cannot be the basis for action. I have argued that
Williams and others are right to insist on this type of 'remainder' but wrong to think
that an argument has been provided against moral cognitivism or realism.
This has, I believe, disposed of Williams's case against moral cognitivism as it was
presented in 'Ethical Consistency'. But we have yet to look at the points made in the
second article, 'Consistency and Realism', which seem to be different.
In 'Consistency and Realism' Williams first contrasts demands for 'consistency' of
beliefs and demands for 'consistency' of imperatives, suggesting that beliefs claim to
represent reality and therefore cannot without error be inconsistent. It cannot be the
case that two inconsistent beliefs are both equally good. This could however be so
with inconsistent imperatives. There are, at most, pragmatic, practical, reasons for
avoiding issuing inconsistent imperatives or other directives; since imperatives are
not about the world. In the final paragraphs of the article he argues that to compare
moral judgements with imperatives in this respect is to give sense to an anti-realist
theory of ethics.
For a realist view would seem to determine a different view of consistency in ethics
from that determined by a non-realist view. . . . On a realist view, the significance of
consistency, and the explanation of it as an aim, are going to come down to the
simple point that moral judgements being straight forwardly assertions, two
inconsistent moral judgements cannot both be true,

end p.52

and hence (truth being the aim of assertions) cannot both be acceptable: one of them
must be rejected; its reasons must be defective; something must be wrong. . . .
But. . . the non-realist approach may well allow for the possibility that one can be

forced to two inconsistent moral judgements about the same situation, each of them
backed by the best possible reasons, and each of them firmly demanding acceptance
. . .

Williams, Problems of the Self, 204-5.

He gives no argument for thinking that this is how it is with moral judgements, simply
referring back to 'Ethical Consistency' for his reasons for favouring a non-realist
theory. As we have found these to be bad reasons the comparison between moral
judgements and imperatives must be considered on its merits. The essence of
Williams's case is, clearly, that there may be two equally acceptable but inconsistent
moral judgements just as there may be two equally acceptable but inconsistent
imperatives; there may be nothing wrong with either of the moral judgements just as
there may be nothing wrong with either imperative, and he goes on to say that if we
do favour moral systems in which such inconsistent judgements do not appear this is
not because moral judgements are straightforward assertions which claim to mirror
reality, but rather because we have pragmatic reasons for doing so. We are to
compare moral judgements to imperatives in order to be able if we so wish to accept
'inconsistent moral judgements about the same situation'.
To understand this argument we must once more ask what Williams means by
"inconsistent" in the expression 'inconsistent moral judgements'; and when we do so
we find that the interpretation of inconsistency is just what the argument is about.
There are two possibilities. The first is that "It is wrong to do a" and "It is wrong not to
do a" are inconsistent in the way that straightforward assertions are inconsistent. The
second is that the propositions are inconsistent only as "Do a" and "Do not do a" are
'inconsistent', that is in their practical aspect, as enjoining inconsistent states of
affairs. On the first interpretation inconsistent moral judgements contain a
contradiction, and cannot both be asserted because reality cannot be such as to
accommodate them both. On the second interpretation the 'inconsistent' moral
judgements merely

end p.53

conflict as 'inconsistent' orders conflict; and just as such orders may be issued so
may conflicting moral judgements be voiced. Although Williams himself does not put
it this way we may see him as arguing that there are moral situations which by
forcing inconsistent moral judgements upon us make it necessary for us to interpret
this inconsistency as conflict rather than contradiction.
What should we think about this argument? The first thing to notice is that it is an
argument that cannot be dealt with in the old way, i.e. by showing that where
Williams sees inconsistent moral judgements we have in truth consistent assertion of
type 1 obligations or oughts. For Williams is not here arguing from the remainder that
exists even where a moral conflict is most clearly resolvable—where it is most
obvious that morally it would be better to do a than not to do a, or the other way
round. He is now arguing from special cases in which no such resolution is in sight.
The presence of this different argument is marked by insistence on the difficulty of
resolving hard cases such as that of Agamemnon's cruel dilemma, an insistence that
would be irrelevant if 'remainder' were all that was in question. It seems clear that this
is how we should interpret the passage in another article, 'Politics and Moral
Character', in which Williams refers to 'cases of tragic choice where one might say
that whatever the agent did was wrong'.

'Politics and Moral Character', 61.

Moreover, his solution to the problem does allow us some understanding of what it
could mean to say that in some situations an agent might be wrong whatever he did,

as being unjustified in whichever course of action he took, in spite of the fact that a
right choice did not exist, and even where fault could not be imputed to him for being
in this situation. It is as if the agent made the wrong choice although the right choice
does not exist. The thought remains puzzling, but the picture can be somewhat filled
in, I suppose, by thinking of a subject to whom conflicting orders are issued, and who
is penalized by the authorities for disobeying one order or for disobeying the other,
whichever it is that he does. Perhaps we sometimes think of ourselves as if we were
in this position vis-à-vis 'the moral law' and are confirmed in this idea by the
discomforts we suffer in certain cases of hard choice.

end p.54

We see then that Williams's new argument against the moral realist—the one that
starts from hard-to-resolve cases—cannot be met in the old way and cannot be
dismissed out of hand as unintelligible.
Nor is there any reason to deny the assumption that there sometimes are cases of
irresolvable conflict between moral principles or values. Bas van Fraassen writes of
irresolvable moral conflict saying 'By this I mean a conflict between what ought to be
for one reason and what ought to be for another reason, which cannot be resolved in
terms of one reason overriding another, or one law or authority or value being higher
than another.'

Van Fraassen, 'Values and the Heart's Command', 8.

And he also refers to Sartre's contention 'that no ethical system can resolve all moral
dilemmas'

Ibid. 10.

and to the possibility that there are cases in which 'our morality's guidance is
ambiguous, inconsistent, or absent altogether'.

Ibid. 11-12.

Marcus writes, in the same vein, 'it would appear that however strong our wills and
complete our knowledge we might be faced with a moral choice in which there are no
moral grounds for favouring doing x over y'.

Marcus, 'Moral Dilemma and Consistency', 125.

I do not think that there is any reason to deny incommensurability of the kind spoken
of by van Fraassen and Marcus, as also by David Wiggins when he writes 'It seems
that in the sphere of the practical we may know for certain that there exist absolutely
undecidable questions—e.g., cases where the situation is so appalling or the choices
are so gruesome that nothing could count as the reasonable practical answer.'

David Wiggins, 'Truth, Invention, and the Meaning of Life', Proceedings of the British Academy, 62

(1976), 371; repr. in Wiggins, Needs, Values, Truth: Essays in the Philosophy of value (Oxford:
Clarendon Press, 1998).

Whether undecidability exists particularly in situations where the alternatives are
ghastly I am not so sure. It certainly isn't the case that there are never better and
worse alternatives here, and it is perhaps particularly important to hang on to this
fact, given the temptation to think 'The choices are all so awful morality really doesn't
tell me what to do' just because the going is especially rough. For all I know there
may be just as much undecidability in small moral matters, or where the choice is
between goods rather than evils, only it doesn't worry us and we don't notice it so
much.

end p.55

It is noticeable that although van Fraassen, Marcus, and Wiggins all make references
to irresolvable moral conflicts they differ from Williams in that they do not seem to
draw any anti-cognitivist conclusion from their observations. The question therefore
arises as to whether Williams has some special assumption which makes him think

such a conclusion irresistible? And whether, if so, his argument can be undermined
by challenging this assumption? The answer to both these questions is affirmative.
Williams must be making the crucial and questionable assumption that in cases of
irresolvable moral dilemma, where the application of one principle would give the
judgement "there is stronger reason morally speaking to do a than to do b"and the
other " . . . to do b than to do a", and there is no way of resolving the issue in favour
of one rather than the other, both judgements have to be affirmed. This must be what
he means when he speaks of both as 'firmly demanding acceptance'. For as I have
suggested the new argument in 'Consistency and Realism'—the one that is based on
the existence of irresolvable moral conflict—depends on the idea that contradictory
affirmation has to be avoided by a shift from the cognitive to the imperatival mode.
And the need arises only if joint affirmation is indeed in question.
What we must ask, therefore, is whether in cases of irresolvable moral conflict we
have to back both the judgement in favour of a and the judgement in favour of b,
although doing b involves not doing a. Is it not possible that we should rather declare
that the two are incommensurable, so that we have nothing to say about the overall
merits of a and b, whether because there is nothing that we can say or because there
is no truth of the matter and therefore nothing to be said? The acceptance of
incommensurability in morality will of course raise many interesting questions, such
as how we recognize it. But incommensurability is not an unfamiliar idea. I think, for
instance, of the impossibility of saying in many cases whether one man is happier
than another when one lives a quiet and contented life and the other a life that is full
of joy and pain. On occasions we make comparisons of happiness with perfect
confidence, and yet sometimes there seems nothing to be said. Perhaps we should
similarly accept incommensurability in certain cases where conflicting moral
judgements suggest themselves to us. And if we do this we do not

end p.56

have to avoid a 'contradiction'between two of our affirmations by interpreting these
affirmations in a special way.
This concludes my argument against Williams. Let me add two observations by way
of postscript. The first is about the interest of the 'remainder' thesis. I should say that
it is not very interesting in so far as it concerns the fact that obligations that are
overridden are not annulled and that there are type 1 propositions about what ought
to be done as well as those of type 2. On the other hand the thesis of 'remainders'in
so far as it is about what is inevitably regrettable in the outcome of moral conflicts is
very interesting indeed. The most interesting part of the topic has, I think, to do with
the inevitable loss involved in a choice between values:when one really good thing
which the man of virtue must cherish has to be sacrificed for another, a loss that is
often reflected in a conflict of oughts or obligations but is not described simply by
talking about such conflicts. It is Sir Isaiah Berlin who has done most to make us
aware of the reality of inevitable loss of one value or another in the political sphere,
as when he says, in 'Two Concepts of Liberty': 'The extent of a man's, or a people's,
liberty to choose to live as they desire must be weighed against the claims of many
other values, of which equality, or justice, or happiness, or security, or public order
are perhaps the most obvious examples.'

Isaiah Berlin, Four Essays on Liberty (Oxford: University Press, 1980), 170.

David Wiggins too has also recently stressed such things and he has been
particularly concerned with the rival goods that a man may see as giving meaning to
his life. In neither author does the full difficulty of the most difficult part of the thought
about inevitable loss emerge. I mean the thought that so far from forming a unity in

the sense that Aristotle and Aquinas believed they did, the virtues actually conflict
with each other: which is to say that if someone has one of them he inevitably fails to
have some other. Many people do not see the difficulty of this idea because they
interpret it rather superficially, as the thought that, e.g., the claims of justice and
charity may conflict. But this is easy to accommodate. For in so far as a man's charity
is limited only by his justice—say the readiness to help someone by his recognition of
this person's right or the right of some other person to non-interference—he

end p.57

is not less than perfect in charity. The far more difficult thought is that he can only
become good in one way by being bad in another, as if, e.g., he could only rein in his
ruthless desires at the cost of a deep malice against himself and the world; or as if a
kind of dull rigidity were the price of refusing to do what he himself wanted at
whatever cost to others. So Nietzsche found thoughts about the possibility that
hatred, envy, covetousness, and the lust to rule must be present in the 'general
economy of life', and must be 'further enhanced if life is further to be enhanced',
terrible thoughts; but with his extraordinary and characteristic courage did not decide
that therefore they must be false. Perhaps we have shied away from such ideas
because we so thoroughly reject Nietzsche. In any case the subject seems a hard
one which stands ready to be explored.
I would also add the following comment which has to do with the distinction between
cognitivism and realism in ethics. I do not know whether one will be able to
distinguish the two unless one understands realism as Michael Dummett does, but if
realism is what Dummett meant by realism then it is obvious that they are distinct.

Michael Dummett, Truth and Other Enigmas (London: Duckworth; Cambridge, Mass.: Harvard

University Press, 1978), esp. preface and pp. 1-24 and 145-65.

A cognitivist theory of ethics says that moral judgements are about the world as are
other kinds of assertions—with, no doubt, many significant variations. A realist theory
of ethics would be one that refused to let the possession of truth or falsity by a
particular ethical proposition stand or fall by our capacity or lack of capacity for
assigning truth or falsity to it. Thus the issue of realism, understood like this, has to
do with the implications of incommensurability. Those who say that where the
solution of some ethical conflicts is beyond our capacities there will nevertheless be a
solution which is perhaps known to God are realists: those who deny it are anti-
realists as regards the class of propositions in which moral judgements are asserted.
I add this merely as a final comment, because Williams seems to have meant by
"realism" not this but rather cognitivism. Whichever way we understand realism I
think his arguments fail.

end p.58

Utilitarianism and the

Virtues

Philippa Foot

It is remarkable how utilitarianism tends to haunt even those of us who will not
believe in it. It is as if we for ever feel that it must be right, although we insist that it is
wrong. T. M. Scanlon hits the nail on the head when he observes, in his article
'Contractualism and Utilitarianism', that the theory occupies a central place in the
moral philosophy of our time in spite of the fact that, as he puts it, 'the implications of
act utilitarianism are wildly at variance with firmly held moral convictions, while rule
utilitarianism . . . strikes most people as an unstable compromise'.

T. M. Scanlon, 'Contractualism and Utilitarianism', in A. Sen and B. Williams (eds.), Utilitarianism and

Beyond

(Cambridge, 1982).

He suggests that what we need to break this spell is to find a better alternative to
utilitarian theories, and I am sure that that is right. But what I want to do is to
approach the business of exorcism more directly. Obviously something drives us
towards utilitarianism, and must it not be an assumption or thought which is in some
way mistaken? For otherwise why is the theory unacceptable? We must be going
wrong somewhere and should find out where it is.
I want to argue that what is most radically wrong with utilitarianism is its
consequentialism, but I also want to suggest that its consequentialist element is one
of the main reasons why utilitarianism seems

end p.59

so compelling. I need therefore to say something about the relation between the two
theory descriptions 'utilitarian' and 'consequentialist'. Consequentialism in its most
general form simply says that it is by 'total outcome', that is, by the whole formed by
an action and its consequences, that what is done is judged right or wrong. A
consequentialist theory of ethics is one which identifies certain states of affairs as
good

states of affairs and says that the rightness or goodness of actions (or of other

subjects of moral judgment) consists in their positive productive relationship to these
states of affairs. Utilitarianism as it is usually defined consists of consequentialism
together with the identification of the best state of affairs with the state of affairs in
which there is most happiness, most pleasure, or the maximum satisfaction of desire.
Strictly speaking utilitarianism—taken here as welfare utilitarianism—is left behind
when the distribution of welfare is said in itself to affect the goodness of states of
affairs; or when anything other than welfare is allowed as part of the good. But it is of
course possible also to count a theory as utilitarian if right action is taken to be that
which produces 'good states of affairs', whatever these are supposed to be; and then
'utilitarianism' becomes synonymous with 'consequentialism'. By 'utilitarianism' I shall
here mean 'welfare utilitarianism', though it is with consequentialism in one form or
another that I shall be most concerned.
Although I believe that what is radically wrong with utilitarianism is its
consequentialism, what has often seemed to be most wrong with it has been either
welfarism or the sum ranking of welfare. So it has been suggested that 'the good' is
not automatically increased by an increase in pleasure, but by non-malicious
pleasure, or first-order pleasure, or something of the kind; in order to get over
difficulties about the pleasures of watching a public execution or the pleasures and
pains of the bigot or the prude.

See e.g. A. Sen, 'Utilitarianism and Welfarism', Journal of Philosophy, 76/9 (1979).

Furthermore, distribution principles have been introduced so that actions benefiting
the rich more than they harm the poor no longer have to be judged morally worthy.
Thus the criteria for the goodness of states of affairs have continually been modified
to meet one objection after another; but it seems that the modifications have never
been able to catch up with the objections. For the distribution

end p.60

principles and the discounting of certain pleasures and pains did nothing to help with
problems about, e.g., the wrongness of inducing cancer in a few experimental
subjects to make a substantial advance in finding a cure for the disease. If the theory
was to give results at all in line with common moral opinion rights had to be looked
after in a way that was so far impossible within even the modified versions of
utilitarianism.

It was therefore suggested, by Amartya Sen, that 'goal rights' systems should be
considered; the idea being that the respecting or violating of rights should be counted
as itself a good or an evil in the evaluation of states of affairs.

A. Sen, 'Rights and Agency', Philosophy and Public Affairs, 11/1 (1981).

This would help to solve some problems because if the respecting of the rights of the
subject were weighted heavily enough the cancer experiment could not turn out to be
'optimific' after all. Yet this seems rather a strange suggestion, because as Samuel
Scheffler has remarked, it is not clear why, in the measurement of the goodness of
states of affairs or total outcomes, killings for instance should count so much more
heavily than deaths.

S. Scheffler, The Rejection of Consequentialism (Oxford,

1982

), 108-12.

But what is more important is that this 'goal rights' system fails to deal with certain
other examples of actions that most of us would want to call wrong. Suppose, for
instance, that some evil person threatens to kill or torture a number of victims unless
we kill or torture one, and suppose that we have every reason to believe that he will
do as he says. Then in terms of their total outcomes (again consisting of the states of
affairs made up of an action and its consequences) we have the choice between
more killings or torturings and less, and a consequentialist will have to say that we
are justified in killing or torturing the one person, and indeed that we are morally
obliged to do it, always supposing that no indirect consequences have tipped the
balance of good and evil. There will in fact be nothing that it will not be right to do to a
perfectly innocent individual if that is the only way of preventing another agent from
doing more things of the same kind.
Now I find this a totally unacceptable conclusion and note that it is a conclusion not of
utilitarianism in particular but rather of consequentialism in any form. So it is the
spellbinding force of consequentialism that we have to think about. Welfarism has its
own peculiar attraction, with has to do with the fact that pleasure, happiness, and the
satisfaction

end p.61

of desire are things seen as in some way good. But this attraction becomes less
powerful as distribution principles are added and pleasures discounted on an ad hoc
basis to destroy the case for such things as public executions.
If having left welfarist utilitarianism behind we still find ourselves unable, in spite of its
difficulties, to get away from consequentialism, there must be a reason for this. What
is it, let us now ask, that is so compelling about consequentialism? It is, I think, the
rather simple thought that it can never be right to prefer a worse state of affairs to a
better.

The original version continued 'How could it ever be right, we think, to produce less good rather than

more good?'I have excised this sentence because in the context the use of the expression 'doing more
good' suggested an identification which I was at pains to deny. At all times I have allowed doing good
as an unproblematic notion, because although it does raise many problems, e.g. about different
distributions of benefits, it does not raise the particular problems with which I am concerned. I want to
insist that however well we might understand what it was to 'do as much good as possible' in the
sense of producing maximum benefit, it would not follow that we knew what we meant by expressions
such as 'the best outcome' or 'the best state of affairs' as these are used by moral philosophers. Cf.
the discussion on pp. 63 ff. of the present version of this paper.

It is this thought that haunts us and, incidentally, this thought that makes the move to
rule utilitarianism an unsatisfactory answer to the problem of reconciling utilitarianism
with common moral opinion. For surely it will be irrational, we feel, to obey even the
most useful rule if in a particular instance we clearly see that such obedience will not
have the best results

. Again following Scheffler we ask if it is not paradoxical that it

should ever be morally objectionable to act in such a way as to minimize morally
objectionable acts of just the same type.

Scheffler, The Rejection of Consequentialism, 121.

If it is a bad state of affairs in which one of these actions is done it will presumably be
a worse state of affairs in which several are. And must it not be irrational to prefer the
worse to the better state of affairs?
This thought does indeed seem compelling. And yet it leads to an apparently
unacceptable conclusion about what it is right to do. So we ought, as I said, to
wonder whether we have not gone wrong somewhere. And I think that indeed we
have. I believe (and this is the main thesis of the paper) that we go wrong in
accepting the idea that there are better and worse states of affairs in the sense that
consequentialism requires. As

end p.62

Wittgenstein says in a different context, 'The decisive movement in the conjuring trick
has been made, and it was the very one that we thought quite innocent.'

L. Wittgenstein, Philosophical Investigations (London, 1953; Oxford, 1958), §308.

Let us therefore look into the idea of a good state of affairs, as this appears in the
thought that we can judge certain states of affairs to be better than others and then
go on to give moral descriptions to actions related productively to these states of
affairs.
We should begin by asking why we are so sure that we even understand expressions
such as 'a good state of affairs' or 'a good outcome'; for as Peter Geach pointed out
years ago there are phrases with the word 'good' in them, as, e.g., 'a good event',
that do not at least as they stand have a sense.

P. Geach, 'Good and Evil', Analysis, 17 (1956).

Following this line one might suggest that philosophers are a bit hasty in using
expressions such as 'a better world'. One may perhaps understand this when it is
taken to mean a 'deontically better world' defined as one in which fewer duties are
left unfulfilled: but obviously this will not help to give a sense to 'better state of affairs'
as the consequentialist needs to use this expression, since he is wanting to fix our
obligations not to refer to their fulfilment.
Nevertheless, it may seem that combinations of words such as 'a good state of
affairs' are beyond reproach or question, for such expressions are extremely familiar.
Do we not use them every day? We say that it is a good thing that something or other
happened; what difficulty can there be in constructing from such elements anything
we want in the way of aggregates such as total outcomes which (in principle) take
into account all the elements of a possible world and so constitute good states of
affairs? Surely no one can seriously suggest that 'good state of affairs' is an
expression that we do not understand?
It would, of course, be ridiculous to query the sense of the ordinary things that we say
about its being 'a good thing' that something or other happened, or about a certain
state of affairs being good or bad. The doubt is not about whether there is some way
of using the words, but rather about the way they appear in the exposition of
utilitarian and other consequentialist moral theories. It is important readily to accept

end p.63

the fact that we talk in a natural and familiar way about good states of affairs, and
that there is nothing problematic about such usage. But it is also important to see
how such expressions actually work in the contexts in which they are at home, and in
particular to ask about the status of a good state of affairs. Is it something impersonal
to be recognized (we hope) by all reasonable men? It seems, surprisingly, that this is

not the case at least in many contexts of utterance of the relevant expressions.
Suppose, for instance, that the supporters of different teams have gathered in the
stadium and that the members of each group are discussing the game: or that two
racegoers have backed different horses in a race. Remarking on the course of events
one or the other may say that things are going well or badly, and when a certain
situation has developed may say that it is a good or a bad state of affairs. More
commonly they will welcome some developments and deplore others, saying 'Oh
good!' or 'That's bad!', calling some news good news and some news bad,
sometimes describing what has happened as 'a good thing' and sometimes not. We
could develop plenty of other examples of this kind, thinking for instance of the
conversations about the invention of a new burglar alarm that might take place in the
police headquarters and in the robbers' den.
At least two types of utterance are here discernible. For 'good' and its cognates may
be used to signal the speaker's attitude to a result judged as an end result, and then
he says 'Good!'or 'I'm glad' or 'That's good' where what he is glad about is something
welcomed in itself and not for any good it will bring. But a state of affairs may rather
be judged by its connection with other things called good. And even what is counted
as in itself good may be said to be bad when it brings enough evil in its train.
Now what shall we say about the truth or falsity of these utterances? It certainly
seems that they can be straightforwardly true or false. For perhaps what appears to
be going to turn out well is really going to turn out badly: what seemed to be a good
thing was really a bad thing, and an apparently good state of affairs was the prelude
to disaster. 'You are quite wrong' one person may say to another and events may
show that he was wrong. Nevertheless, we can see that this quasi-objectivity, which
is not to be questioned when people with similar aims, interests, or desires are

end p.64

speaking together, flies out of the window if we try to set the utterances of those in
one group against the utterances of those in another. One will say 'a good thing'
where another says 'a bad thing', and it is the same for states of affairs. It would be
bizarre to suggest that at the races it really is a good thing that one horse or the other
is gaining (perhaps because of the pleasure it will bring to the majority, or the good
effect on the future of racing) and so that the utterance of one particular punter, intent
only on making a packet, will be the one that is true.
This is not to say, however, that what a given person says to be a good thing or a
good state of affairs must relate to his own advantage. For anyone may be interested
in

the future of racing, and people commonly are interested in, e.g., the success of

their friends, saying 'that's a good thing' if one of them looks like winning a prize or
getting a job; incidentally without worrying much about whether he is the very best
candidate for it.
Now it may be thought that these must be rather special uses of expressions such as
'good state of affairs', because we surely must speak quite differently when we are
talking about public matters, as when for instance we react to news of some far-away
disaster. We say that the news is bad because a lot of people have lost their lives in
an earthquake. Later we may say that things are not as bad as we feared and
someone may remark 'that's a good thing', 'A bad state of affairs', we might remark
on hearing the original news about people dead or homeless, and this will usually
have nothing to do with harm to us or to our friends.
In this way the case is different from that of the racegoers or the cops and robbers,
but this is not of course to imply that what we say on such occasions has a different
status from the utterances we have considered so far. For why should its truth not be

'speaker-relative' too, also depending on what the speakers and their group are
interested in

though not now on the good or harm that will come to them themselves?

Is it not more plausible to think this than to try to distinguish two kinds of uses of
these expressions, one speaker-relative and the other not? For are there really two
ways in which the police for instance might speak? And two ways in which the
robbers could speak as well? Are we really to say that although when they are both
speaking in the speaker-relative way

end p.65

they do not contradict each other, and may both speak truly, when speaking in the
'objective' way one group will speak truly and the other not? What shows that the
second way of speaking exists?
What thoughts, one may ask, can we really be supposed to have which must be
expressed in the disputed mode? Considering examples such as that of the far-away
earthquake we may think that we believe the best state of affairs to be the one in
which there is most happiness and least misery, or something of the sort. But
considering other examples we may come to wonder whether any such thought can
really be attributed to us.
Suppose for instance that when walking in a poor district one of us should lose a
fairly considerable sum of money which we had intended to spend on something
rather nice. Arriving home we discover the loss and telephone the police on the off
chance that our wad of notes has been found and turned in. To our delight we find
that it was picked up by a passing honest policeman, and that we shall get it back.
'What a good thing' we say 'that an officer happened to be there. 'What seemed to be
a bad state of affairs has turned out not to be bad after all: things are much better
than we thought they were. And all's well that ends well. But how, it may now be
asked, can we say that things have turned out better than we thought? Were we not
supposed to believe that the best state of affairs was the one in which there was
most happiness and least misery? So surely it would have been better if the money
had not been returned to us but rather found and kept as treasure trove by some
poor inhabitant of the region? We simply had not considered that because most of us
do not actually have the thought that the best state of affairs is the one in which we
lose and they gain. Perhaps we should have had this thought if it had been a small
amount of money, but this was rather a lot.
No doubt it will seem to many that there must be non-speaker-relative uses of words
evaluating states of affairs because moral judgements cannot have speaker-relative
status. But if one is inclined, as I am, to doubt whether propositions of this form play
any part in the fundamentals of ethical theory there is no objection on this score. It is
important however that the preceding discussion has been about propositions of a
particular form and nothing has been said to suggest that all judgements about what
is good and bad have speaker-relative status. I have not for

end p.66

instance made this suggestion for what Geach called 'attributive' judgements
concerning things good or bad of a kind—good knives and houses and essays, or
even good actions, motives, or men. If there is some reason for calling these
'speaker-relative' the reason has not been given here. Nor has anything been said
about the status of propositions about what is good for anyone or anything, or about
that in which their good consists.
What has I hope now been shown is that we should not take it for granted that we
even know what we are talking about if we enter into a discussion with the
consequentialist about whether it can ever be right to produce something other than
'the best state of affairs'.

It might be suggested by way of reply that what is in question in these debates is not
just the best state of affairs without qualification but rather the best state of affairs
from an impersonal point of view

. But what does this mean? A good state of affairs

from an impersonal point of view is presumably opposed to a good state of affairs
from my point of view or from your point of view, and as a good state of affairs from
my point of view is a state of affairs which is advantageous to me, and a good state
of affairs from your point of view is a state of affairs that is advantageous to you, a
good state of affairs from an impersonal point of view presumably means a state of
affairs which is generally advantageous, or advantageous to most people, or
something like that. About the idea of maximum welfare we are not (or so we are
supposing for the sake of the argument) in any difficulty.

Cf. n. 5.

But an account of the idea of a good state of affairs which simply defines it in terms
of maximum welfare is no help to us here. For our problem is that something is
supposed to be being said about maximum welfare and we cannot figure out what
this is.
In a second reply, more to the point, the consequentialist might say that what we
should really be dealing with in this discussion is states of affairs which are good or
bad, not simply, but from the moral point of view. The qualification is, it will be
suggested, tacitly understood in moral contexts, where no individual speaker gives
his own private interests or allegiances a special place in any debate, the speaker-
relativity

end p.67

found in other contexts thus being left behind. This seems to be a pattern familiar
from other cases, as, e.g., from discussions in meetings of the governors of public
institutions. Why should it not be in a similar way that we talk of a good and a bad
thing to happen 'from a moral point of view'? And is it not hard to reject the
conclusion that right action is action producing this 'best state of affairs'?
That special contexts can create special uses of the expressions we are discussing is
indeed true. But before we proceed to draw conclusions about moral judgements we
should ask why we think that it makes sense to talk about morally good and bad
states of affairs, or to say that it is a good thing (or is good that) something happened
'from a moral point of view'. For after all we cannot concoct a meaningful sentence by
adding just any qualification of this verbal form to expressions such as these. What
would it mean, for instance, to say that a state of affairs was good or bad 'from a
legal point of view' or 'from the point of view of etiquette'? Or that it was a good thing
that a certain thing happened from these same 'points of view'? Certain
interpretations that suggest themselves are obviously irrelevant, as, for instance, that
it is a good state of affairs from a legal point of view when the laws are clearly stated,
or a good state of affairs from the point of view of etiquette when everyone follows
the rules.
It seems, therefore, that we do not solve the problem of the meaning of 'best state of
affairs' when supposed to be used in a non-speaker-relative way simply by tacking on
'from a moral point of view'; since it cannot be assumed that the resulting expression
has any sense. Nevertheless, it would be wrong to suggest that 'good state of affairs
from a moral point of view' is a concatenation of words which in fact has no meaning
in any of the contexts in which it appears, and to see this we have only to look at
utilitarian theories of the type put forward by John C. Harsanyi and R. M. Hare, in
which a certain interpretation is implicitly provided for such expressions.

See e.g. J. C. Harsanyi, 'Morality and the Theory of Rational Behavior', Social Research, 44/4

(Winter 1977), repr. in Sen and Williams, Utilitarianism and Beyond; and R. M. Hare, Moral Thinking
(Oxford, 1981).

Harsanyi for instance argues that the only rational morality is one in

end p.68

which the rightness or wrongness of an action is judged by its relation to a certain
outcome, i.e. the maximization of social utility. The details of this theory, which
defines social utility in terms of individual preferences, do not concern us here. The
relevant point is that within it there appears the idea of an end which is the goal of
moral action, and therefore the idea of a best state of affairs from a moral point of
view. (It does not of course matter whether Harsanyi uses these words.)
Similarly Hare, by a more elaborate argument from the universalizability and
prescriptivity of moral judgements, tries to establish the proposition that one who
takes the moral point of view must have as his aim the maximization of utility,
reflecting this in one way in his day-today prescriptions and in another in 'critical'
moral judgements. So here too a clear sense can be given to the idea of a best state
of affairs from a moral point of view: it is the state of affairs which a man aims at
when he takes the moral point of view and which in one way or another determines
the truth of moral judgements.
Within these theories there is, then, no problem about the meaning of expressions
such as 'the best state of affairs from the moral point of view'. It does not follow,
however, that those who reject the theories should be ready to discuss the pros and
cons of consequentialism in these terms. For unless the arguments given by Hare
and Harsanyi are acceptable it will not have been shown that there is any reference
for expressions such as 'the aim which each man has in so far as he takes up the
moral point of view' or a fortiori 'the best state of affairs from the moral point of view'.
If my main thesis is correct this is a point of the first importance. For I am arguing that
where non-consequentialists commonly go wrong is in accepting from their
opponents questions such as 'Is it ever right to act in such a way as to produce
something less than the best state of affairs that is within one's reach?'

See e.g. T. Nagel, 'The Limits of Objectivity', in S. McMursin (ed.), The Tanner Lectures on Human

Values

, vol. i (1980). 131, where he says that things would be better, what happened would be better'

if I twisted a child's arm in circumstances where (by Nagel's hypothesis) this was the only way to get
medical help for the victims of an accident. He supposes that I might have done something worse if I
hurt the child than if I did not do it, but that the total outcome would have been better. It does not, I
think, occur to him to question the idea of things being better—or things being worse.

Summing up the results reached so far we may

end p.69

say that if taken in one way, with no special reference to morality, talk about good
states of affairs seems to be speaker-relative. But if the qualification 'from a moral
point of view' is added the resulting expression may mean nothing; and it may lack a
reference when a special consequentialist theory has given it a sense.
In the light of this discussion we should find it significant that many people who do
not find any particular consequentialist theory compelling nevertheless feel
themselves driven towards consequentialism by a thought which turns on the idea
that there are states of affairs which are better or worse from a moral point of view.
What is it that seems to make this an inescapable idea?
Tracing the assumption back in my own mind I find that what seems preposterous is
to deny that there are some things that a moral person must want and aim at in so far
as he is a moral person and that he will count it 'a good thing' when these things
happen and 'a good state of affairs' either when they are happening or when things

are disposed in their favour. For surely he must want others to be happy. To deny
this would be to deny that benevolence is a virtue—and who wants to deny that?
Let us see where this line of thought will take us, accepting without any reservation
that benevolence is a virtue and that a benevolent person must often aim at the good
of others and call it 'a good thing' when for instance a far-away disaster turns out to
have been less serious than was feared. Here we do indeed have the words 'a good
thing'(and just as obviously a 'good state of affairs') necessarily appearing in moral
contexts. And the use is explained not by a piece of utilitarian theory but by a simple
observation about benevolence.
This, then, seems to be the way in which seeing states of affairs in which people are
happy as good states of affairs really is an essential part of morality. But it is very
important that we have found this end within morality, and forming part of it, not
standing outside it as the 'good state of affairs' by which moral action in general is to
be judged. For benevolence is only one of the virtues, and we shall have to look at
the others before we can pronounce on any question about good or bad action in
particular circumstances. Offhand we have no reason to think that

end p.70

whatever is done with the aim of improving the lot of other people will be morally
required or even morally permissible. For firstly there are virtues such as friendship
which play their part in determining the requirements of benevolence, e.g. by making
it consistent with benevolence to give service to friends rather than to strangers or
acquaintances. And secondly there is the virtue of justice, taken in the old wide
sense in which it had to do with everything owed. In our common moral code we find
numerous examples of limitations which justice places on the pursuit of welfare. In
the first place there are principles of distributive justice which forbid, on grounds of
fairness, the kind of 'doing good' which increases the wealth of rich people at the cost
of misery to the poor. Secondly, rules such as truth-telling are not to be broken
wherever and whenever welfare would thereby be increased. Thirdly, considerations
about rights, both positive and negative, limit the action which can be taken for the
sake of welfare. Justice is primarily concerned with the following of certain rules of
fairness and honest dealing and with respecting prohibitions on interference with
others, rather than with attachment to any end. It is true that the just man must also
fight injustice, and here justice like benevolence is a matter of ends, but of course the
end is not the same end as the one that benevolence seeks and need not be
coincident with it.
I do not mean to go into these matters in detail here, but simply to point out that we
find in our ordinary moral code many requirements and prohibitions inconsistent with
the idea that benevolence is the whole of morality. From the point of view of the
present discussion it would be acceptable to describe the situation in terms of a
tension between, for instance, justice and benevolence. But it is not strictly accurate
to think of it like this, because that would suggest that someone who does an unjust
act for the sake of increasing total happiness has a higher degree of benevolence
than one who refuses to do it. Since someone who refuses to sacrifice an innocent
life for the sake of increasing happiness is not to be counted as less benevolent than
someone who is ready to do it, this cannot be right. We might be tempted to think
that the latter would be acting 'out of benevolence' because his aim is the happiness
of others, but this seems a bad way of talking. Certainly benevolence does not

end p.71

require unjust action, and we should not call an act which violated rights an act of
benevolence. It would not, for instance, be an act of benevolence to induce cancer in

one person (or deliberately to let it run its course) even for the sake of alleviating
much suffering.
What we should say therefore is that even perfection in benevolence does not imply
a readiness to do anything and everything of which it can be said that it is highly
probable that it will increase the sum of human happiness. And this, incidentally,
throws some light on a certain type of utilitarian theory which identifies the moral
assessment of a situation with that of a sympathetic impartial observer whose
benevolence extends equally to all mankind.

See Harsanyi, 'Morality and the Theory of Rational Behavior'; Sen and Williams, Utilitarianism and

Beyond

, 39.

For what, we may ask, are we to suppose about this person's other characteristics?
Is he to be guided simply and solely by a desire to relieve suffering and increase
happiness; or is he also just? If it is said that for him the telling of truth, keeping of
promises, and respecting of individual autonomy are to be recommended only in so
far as these serve to maximize welfare then we see that the 'impartial sympathetic
observer' is by definition one with a utilitarian point of view. So the utilitarians are
defining moral assessment in their own terms.
Returning to the main line of our argument we now find ourselves in a better position
to see that there indeed is a place within morality for the idea of better and worse
states of affairs. That there is such a place is true if only because the proper end of
benevolence is the good of others, and because in many situations the person who
has this virtue will be able to think of good and bad states of affairs, in terms of the
general good. It does not, however, follow that he will always be able to do so. For
sometimes justice will forbid a certain action, as it forbids the harmful experiment
designed to further cancer research; and then it will not be possible to ask whether
'the state of affairs' containing the action and its results will be better or worse than
one in which the action is not done. The action is one that cannot be done, because
justice forbids it, and nothing that has this moral character comes within the scope of
the kind of comparison of total outcomes that benevolence may sometimes require.
Picking up at this point the example discussed earlier about the

end p.72

morality of killing or torturing to prevent more killings or torturings we see the same
principle operating here. If it were a question of riding out to rescue a small number
or a large number then benevolence would, we may suppose, urge that the larger
number be saved. But if it is a matter of preventing the killing by killing (or conniving
at a killing) the case will be quite different. One does not have to believe that all rights
to noninterference are absolute to believe that this is an unjust action, and if it is
unjust the moral man says to himself that he cannot do it and does not include it in an
assessment he may be making about the good and bad states of affairs that he can
bring about.
What has been said in the last few paragraphs is, I suggest, a sketch of what can
truly be said about the important place that the idea of maximum welfare has in
morality. It is not that in the guise of 'the best outcome' it stands outside morality as
its foundation and arbiter, but rather that it appears within morality as the end of one
of the virtues.
When we see it like this, and give expressions such as 'best outcome' and 'good
state of affairs' no special meaning in moral contexts other than the one that the
virtues give them, we shall no longer think the paradoxical thought that it is
sometimes right to act in such a way that the total outcome, consisting of one's action
and its results, is less good than some other accessible at the time. In the abstract a

benevolent person must wish that loss and harm should be minimized. He does not,
however, wish that the whole consisting of a killing to minimize killings should be
actualized either by his agency or that of anyone else. So there is no reason on this
score to think that he must regard it as 'the better state of affairs'.

I have discussed examples of this kind in more detail in 'Morality, Action, and Outcome', in T.

Honderich (ed.), Objectivity and Value: Essays in Memory of John Mackie (London,

1985

And therefore there is no reason for the non-consequentialist, whose thought of good
and bad states of affairs in moral contexts comes only from the virtues themselves, to
describe the refusal as a choice of a worse total outcome. If he does so describe it he
will be giving the words the sense they have in his opponents' theories, and it is not
surprising that he should find himself in their hands.
We may also remind ourselves at this point that benevolence is not the only virtue
which has to do, at least in part, with ends rather than with

end p.73

the observance rules. As mentioned earlier there belongs to the virtue of justice the
readiness to fight for justice as well as to observe its laws; and there belongs to
truthfulness not only the avoidance of lying but also that other kind of attachment to
truth which has to do with its preservation and pursuit. A man of virtue must be a
lover of justice and a lover of truth. Furthermore, he will seek the special good of his
family and friends. Thus there will be many things which he will want and will
welcome, sometimes sharing these aims with others and sometimes opposing them,
as when working differentially for his own children or his own friends.

See D. Parfit, 'Prudence, Morality, and the Prisoner's Dilemma', Proceedings of the British

Academy

, 65 (1979), and Sen. 'Rights and Agency'.

Similarly someone who is judging a competition and is a fair judge must try to see to
it that the best man wins. The existence of these 'moral aims' will of course give
opportunity for the use, in moral contexts, of such expressions as 'a good thing' or
'the best state of affairs'. But nothing of a consequentialist nature follows from such
pieces of usage, found here and there within morality.
An analogy will perhaps help to make my point. Thinking about good manners we
might decide that someone who has good manners tries to avoid embarrassing
others in social situations. This must, let us suppose, be one of his aims; and we
might even decide that so far as manners is concerned this, or something like it, is
the only prescribed end. But of course this does not mean that what good manners
require of anyone is universally determined by this end. A consequentialist theory of
good manners would presumably be mistaken; because good manners, not being
solely a matter of purposes, also require that certain things be done or not done: e.g.
that hospitality not be abused by frank discussion of the deficiencies of one's host as
soon as he leaves the room. So if invited to take part in such discussions a well-
mannered person will, if necessary, maintain a silence embarrassing to an
interlocutor, because the rule here takes precedence over the aim prescribed.
Assuming that this is a correct account of good manners—and it does not of course
matter whether it is or not—we can now see the difficulty that arises if we try to say
which choice open to the agent results in the best state of affairs from the point of
view of manners. In certain contexts the state of affairs containing no

end p.74

embarrassment will be referred to as a good state of affairs, because avoiding
embarrassment is by our hypothesis the one end prescribed by good manners. But
we should not be surprised if the right action from the point of view of good manners
is sometimes the one that produces something other than this good state of affairs.
We have no right to take an end from within the whole that makes up good manners

and turn it, just because it is an end, into the single guide to action to be used by the
well-mannered man.
This analogy serves to illustrate my point about the illegitimacy of moving what is
found within morality to a criterial position outside it. But it may also bring to the
surface a reason many will be ready to give for being dissatisfied with my thesis. For
surely a morality is unlike a code of manners in claiming rational justification for its
ordinances? It cannot be enough to say that we do have such things as rules of
justice in our present system of virtues: the question is whether we should have
them, and if so why we should. And the reason this is crucial in the present context is
that the justification of a moral code may seem inevitably to involve the very idea that
has been called in question in this paper.
This is a very important objection. In its most persuasive form it involves a picture of
morality as a rational device developed to serve certain purposes, and therefore
answerable to these purposes. Morality, it will be suggested, is a device with a
certain object, having to do with the harmonizing of ends or the securing of the
greatest possible general good, or perhaps one of these things plus the safeguarding
of rights. And the content of morality—what really is right and wrong—will be thought
to be determined by what it is rational to require in the way of conduct given that
these are our aims. Thus morality is thought of as a kind of tacit legislation by the
community, and it is, of course, significant that the early utilitarians, who were much
interested in the rationalizing of actual Parliamentary legislation, were ready to talk in
these terms.

See e.g. J. Bentham, An Introduction to the Principles of Legislation (1789), ch.

, sect. 1.

In moral legislation our aim is, they thought, the general good. With this way of
looking at morality there reappears the idea of better and worse states of affairs from
the moral point of view. Moreover,

end p.75

consequentialism in some form is necessarily reinstated. For while there is room on
such a model for rational moral codes which enjoin something other than the pursuit
of 'the best state of affairs from the moral point of view' this will be only in so far as it
is by means of such ordinances that the object of a moral code is best achieved.

For discussions of this possibility see e.g. R. Adams, 'Motive Utilitarianism', Journal of Philosophy,

73 (1976), and D. Parfit, Reasons and Persons (New York, 1986), 24-8.

Thus it may seem that we must after all allow that the idea of a good state of affairs
appears at the most basic level in the critical appraisal of any moral code. This
would, however, be too hasty a conclusion. Consequentialism in some form follows
from the premiss that morality is a device for achieving a certain shared end. But why
should we accept this view of what morality is and how it is to be judged? Why
should we not rather see that as itself a consequentialist assumption, which has
come to seem neutral and inevitable only in so far as utilitarianism and other forms of
consequentialism now dominate moral philosophy?
To counter this bewitchment let us ask awkward questions about who is supposed to
have

the end which morality is supposed to be in aid of. J. S. Mill notoriously found it

hard to pass from the premiss that the end of each is the good of each to the
proposition that the end of all is the good of all.

J. S. Mill, Utilitarianism (1863), ch.

Perhaps no such shared end appears in the foundations of ethics, where we may
rather find individual ends and rational compromises between those who have them.
Or perhaps at the most basic level lie facts about the way individual human beings
can find the greatest goods which they are capable of possessing. The truth is, I

think, that we simply do not have a satisfactory theory of morality, and need to look
for it. Scanlon was indeed right in saying that the real answer to utilitarianism
depends on progress in the development of alternatives. Meanwhile, however, we
have no reason to think that we must accept consequentialism in any form. If the
thesis of this paper is correct we should be more alert than we usually are to the
possibility that we may unwittingly, and unnecessarily, surrender to consequentialism
by uncritically accepting its key idea. Let us remind ourselves that the idea of the
goodness of total states of affairs played no part in Aristotle's moral philosophy, and
that

end p.76

in modern times in plays no part either in Rawls's account of justice or in the theories
of more thoroughgoing contractualists such as Scanlon.

J. Rawls, A Theory of Justice (Cambridge, Mass., 1971): Scanlon, 'Contractualism and

Utilitarianism'.

If we accustom ourselves to the thought that there is simply a blank where
consequentialists see 'the best state of affairs' we may be better able to give other
theories the hearing they deserve.

end p.77

Killing and Letting

Die

Philippa Foot

Is there a morally relevant distinction between killing and allowing to die? Many
philosophers say that there is not, and further insist that there is no other closely
related difference, as for instance that dividing act from omission, which ever plays a
part in determining the moral character of an action. James Rachels has argued this
case in his well-known article on active and passive euthanasia, Michael Tooley has
argued it in his writings on abortion, and Jonathan Bennett argued it in the Tanner
Lectures given in Oxford in 1980.

James Rachels, 'Active and Passive Euthanasia', New England Journal of Medicine, 292 (9 Jan.

1975); Michael Tooley, 'Abortion and Infanticide', Philosophy and Public Affairs, 2/1 (Fall 1972).
Jonathan Bennett, 'Morality and Conscience', in S. McMurrin (ed.), The Tanner Lectures on Human
Values

, vol. ii (Cambridge: Cambridge University Press, 1981).

I believe that these people are mistaken, and this is what I shall try to show in this
essay. I shall first consider the question in abstraction from any particular practical
moral problem, and then I shall examine the implications my thesis may have
concerning the issue of abortion.
The question with which we are concerned has been dramatically posed by asking
whether we are as much to blame for allowing people in Third World countries to
starve to death as we would be for killing them by sending poisoned food? In each
case it is true that if we acted differently—by sending good food or by not sending
poisoned food—those who are going to die because we do not send the good food or
do send the

end p.78

poisoned food would not die after all. Our agency plays a part in what happens
whichever way they die. Philosophers such as Rachels, Tooley, and Bennett
consider this to be all that matters in determining our guilt or innocence. Or rather
they say that although related things are morally relevant, such as our reasons for
acting as we do and the cost of acting otherwise, these are only contingently related
to the distinction between doing and allowing. If we hold them steady and vary only
the way in which our agency enters into the matter, no moral differences will be

found. It is of no significance, they say, whether we kill others or let them die, or
whether they die by our act or our omission. Although these latter differences may at
first seem to affect the morality of action, we shall always find on further enquiry that
some other difference—such as a difference of motive or cost—has crept in.
Now this, on the face of it, is extremely implausible. We are not inclined to think that it
would be no worse to murder to get money for some comfort such as a nice winter
coat than it is to keep the money back before sending a donation to Oxfam or Care.
We do not think that we might just as well be called murderers for one as for the
other. And there are a host of other examples which seem to make the same point.
We may have to allow one person to die if saving him would mean that we could not
save five others, as for instance when a drug is in short supply and he needs five
times as much as each of them, but that does not mean that we could carve up one
patient to get 'spare parts'for five.
These moral intuitions stand clearly before us, but I do not think it would be right to
conclude from the fact that these examples all seem to hang on the contrast between
killing and allowing to die that this is precisely the distinction that is important from
the moral point of view. For example, having someone killed is not strictly killing him,
but seems just the same morally speaking; and on the other hand, turning off a
respirator might be called killing, although it seems morally indistinguishable from
allowing to die. Nor does it seem that the difference between "act" and "omission"is
quite what we want, in that a respirator that had to be turned on each morning would
not change the moral problems that arise with the ones we have now. Perhaps there
is no locution in the language which exactly serves our purposes and we should
therefore invent

end p.79

our own vocabulary. Let us mark the distinction we are after by saying that one
person may or may not be "the agent"of harm that befalls someone else.
When is one person "the agent"in this special sense of someone else's death, or of
some harm other than death that befalls him? This idea can easily be described in a
general way. If there are difficulties when it comes to detail, some of these ideas may
be best left unsolved, for there may be an area of indefiniteness reflecting the
uncertainty that belongs to our moral judgements in some complex and perhaps
infrequently encountered situations. The idea of agency, in the sense that we want,
seems to be composed of two subsidiary ideas. First, we think of particular effects as
the result of particular sequences, as when a certain fatal sequence leads to
someone's death. This idea is implied in coroners'verdicts telling us what someone
died of, and this concept is not made suspect by the fact that it is sometimes
impossible to pick out a single fatal sequence—as in the lawyers'example of the man
journeying into the desert who had two enemies, one of whom bored a hole in his
water barrel while the other filled it with brine. Suppose such complications absent.
Then we can pick out the fatal sequence and go on to ask who initiated it. If the
subject died by poisoning and it was I who put the poison into his drink, then I am the
agent of his death; likewise if I shot him and he died of a bullet wound. Of course
there are problems about fatal sequences which would have been harmless but for
special circumstances, and those which although threatening would have run out
harmlessly but for something that somebody did. But we can easily understand the
idea that a death comes about through our agency if we send someone poisoned
food or cut him up for spare parts, but not (ordinarily) if we fail to save him when he is
threatened by accident or disease. Our examples are not problem cases from this
point of view.

Nor is it difficult to find more examples to drive our original point home, and show that
it is sometimes permissible to allow a certain harm to befall someone, although it
would have been wrong to bring this harm on him by one's own agency, by
originating or sustaining the sequence which brings the harm. Let us consider, for
instance, a pair of cases which I shall call Rescue I and Rescue II. In the first Rescue
story we are hurrying

end p.80

in our jeep to save some people—let there be five of them—who are imminently
threatened by the ocean tide. We have not a moment to spare, so when we hear of a
single person who also needs rescuing from some other disaster we say regretfully
that we cannot rescue him, but must leave him to die. To most of us this seems clear,
and I shall take it as clear, ignoring John Taurek's interesting if surprising argument
against the obligation to save the greater number when we can.

John Taurek, 'Should the Numbers Count?', Philosophy and Public Affairs, 4 (Summer

1977

This is Rescue I and with it I contrast Rescue II. In this second story we are again
hurrying to the place where the tide is coming in in order to rescue the party of
people, but this time it is relevant that the road is narrow and rocky. In this version
the lone individual is trapped (do not ask me how) on the path. If we are to rescue the
five we would have to drive over him. But can we do so? If we stop he will be all right
eventually: he is in no danger unless from us; but of course all five of the others will
be drowned. As in the first story our choice is between a course of action which will
leave one man dead and five alive at the end of the day and a course of action which
will have the opposite result. And yet we surely feel that in one case we can rescue
the five men and in the other we cannot. We can allow someone to die of whatever
threatens him if the cost of saving him is failing to save five; we cannot, however,
drive over him in order to get to them. We cannot originate a fatal sequence,
although we can allow one to run its course. Similarly, in the pair of examples
mentioned earlier, we find a contrast between on the one hand refusing to give to
one man the whole supply of a scarce drug, because we can use portions of it to
save five, and on the other, cutting him up for spare parts. And we notice that we
may not originate a fatal sequence even if the resulting death is in no sense our
object. We could not knowingly subject one person to deadly fumes in the process of
manufacturing some substance that would save many, even if the poisoning were a
mere side effect of the process that saved lives.
Considering these examples, it is hard to resist the conclusion that it makes all the
difference whether those who are going to die if we act in a certain way will die as a
result of a sequence that we originate or of

end p.81

one that we allow to continue, it being of course something that did not start by our
agency. So let us ask how this could be? If the distinction—which is roughly that
between killing and allowing to die—is morally relevant, because it sometimes makes
the difference between what is right and what is wrong, how does this work? After all,
it cannot be a magical difference, and it does not satisfy anyone to hear that what we
have is just an ultimate moral fact. Moreover, those who deny the relevance can
point to some cases in which it seems to make no difference to the goodness or
badness of an action having a certain result, as, for example, that some innocent
person dies, whether that is due to a sequence we originate or one we merely allow.
And if the way the result comes about sometimes makes no difference, how can it
ever do so? If it sometimes makes an action bad that harm came to someone else as
a result of a sequence we originated, must this not always contribute some element

of badness? How can a consideration be a reason for saying that an action is bad in
one place without being at least a reason for saying the same elsewhere?
Let us address these questions. As to the route by which considerations of agency
enter the process of moral judgement, it seems to be through its connection with
different types of rights. For there are rights to non-interference, which form one class
of rights; and there are rights to goods or services, which are different. And
corresponding to these two types of rights are, on the one hand, the duty not to
interfere, called a "negative duty", and on the other the duty to provide the goods or
services, called a "positive duty". These rights may in certain circumstances be
overridden, and this can in principle happen to rights of either kind. So, for instance,
in the matter of property rights, others have in ordinary circumstances a duty not to
interfere with our property, though in exceptional circumstances the right is
overridden, as in Elizabeth Anscombe's example of destroying someone's house to
stop the spread of a fire.

G. E. M. Anscombe, 'Modern Moral Philosophy', Philosophy, 33 (1958).

And a right to goods or services depending, for example, on a promise will quite often
be overridden in the same kind of case. There is, however, no guarantee that the
special circumstances

end p.82

that allow one kind of right to be overridden will always allow the overriding of the
other. Typically, it takes more to justify an interference than to justify the withholding
of goods or services; and it is, of course, possible to think that nothing whatsoever
will justify, for example, the infliction of torture or the deliberate killing of the innocent.
It is not hard to see how all this connects with the morality of killing and allowing to
die—and in general with harm which an agent allows to happen and harm coming
about through his agency, in my special sense having to do with originating or
sustaining harmful sequences. For the violation of a right to non-interference consists
in interference, which implies breaking into an existing sequence and initiating a new
one. It is not usually possible, for instance, to violate that right to noninterference,
which is at least part of what is meant by "the right to life", by failing to save someone
from death. So if, in any circumstances, the right to non-interference is the only right
that exists, or if it is the only right special circumstances have not overridden, then it
may not be permissible to initiate a fatal sequence, but it may be permissible to
withhold aid.
The question now is whether we ever find cases in which the right to non-interference
exists and is not overridden, but where the right to goods or services either does not
exist or is here overridden. The answer is, of course, that this is quite a common
case. It often happens that whereas someone's rights stand in the way of our
interference, we owe him no service in relation to that which he would lose if we
interfered. We may not deprive him of his property, though we do not have to help
him secure his hold on it, in spite of the fact that the balance of good and evil in the
outcome (counting his loss or gain and the cost to us) will be the same regardless of
how they come about. Similarly, where the issue is one of life and death, it is often
impermissible to kill someone—although special circumstances having to do with the
good of others make it permissible, or even required, that we do not spend the time
or resources needed to save his life, as, for instance, in the story of Rescue I, or in
that of the scarce drug.
It seems clear, therefore, that there are circumstances in which it makes all the
difference, morally speaking, whether a given balance of

end p.83

good and evil came about through our agency (in my sense), or whether it was rather
something we had the ability to prevent but, for good reasons, did not prevent. Of
course we often have a strict duty to prevent harm to others, or to ameliorate their
condition. And even where they do not, strictly speaking, have a right to our goods or
services, we should often be failing (and sometimes grossly failing) in charity if we
did not help them. But, to reiterate, it may be right to allow one person to die in order
to save five, although it would not be right to kill him to bring the same good to them.
How is it, then, that anyone has ever denied this conclusion, so sympathetic to our
everyday moral intuitions and apparently so well grounded in a very generally
recognized distinction between different types of rights? We must now turn to an
argument first given by James Rachels, and more or less followed by others who
think as he does. Rachels told a gruesome story of a child drowned in a bathtub in
two different ways: in one case someone pushed the child's head under water, and in
the other he found the child drowning and did not pull him out. Rachels says that we
should judge one way of acting as bad as the other, so we have an example in which
killing is as bad as allowing to die. But how, he asks, can the distinction ever be
relevant if it is not relevant here?

Rachels, 'Active and Passive Euthanasia'.

Based on what has been said earlier, the answer to Rachels should be obvious. The
reason why it is, in ordinary circumstances, 'no worse' to leave a child drowning in a
bathtub than to push it under is that both charity and the special duty of care that we
owe to children give us a positive obligation to save them, and we have no particular
reason to say that it is 'less bad' to fail in this than it is to be in dereliction of the
negative duty by being the agent of harm. The level of badness is, we may suppose,
the same, but because a different kind of bad action has been done, there is no
reason to suppose that the two ways of acting will always give this same result. In
other circumstances one might be worse than the other, or only one might be bad.
And this last result is exactly what we find in circumstances that allow a positive but
not a negative duty to be overridden.

end p.84

Thus, it could be right to leave someone to die by the roadside in the story of Rescue
I, though wrong to run over him in the story of Rescue II; and it could be right to act
correspondingly in the cases of the scarce drug and the 'spare parts'.
Let me now consider an objection to the thesis I have been defending. It may be said
that I shall have difficulty explaining a certain range of examples in which it seems
permissible, and even obligatory, to make an intervention which jeopardizes people
not already in danger in order to save others who are. The following case has been
discussed. Suppose a runaway tram is heading towards a track on which five people
are standing, and that there is someone who could switch the points, thereby
diverting the tram onto a track on which there is only one person. It seems that he
should do this, just as a pilot whose plane is going to crash has a duty to steer, if he
can, towards a less crowded street than the one he sees below. But the railway man
then puts the one man newly in danger, instead of allowing the five to be killed. Why
does not the one man's right to non-interference stand in his way, as one person's
right to noninterference impeded the manufacture of poisonous fumes when this was
necessary to save five?
The answer seems to be that this is a special case, in that we have here the diverting
of a fatal sequence and not the starting of a new one. So we could not start a flood to
stop a fire, even when the fire would kill more than the flood, but we could divert a
flood to an area in which fewer people would be drowned.

A second and much more important difficulty involves cases in which it seems that
the distinction between agency and allowing is inexplicably irrelevant. Why, I shall be
asked, is it not morally permissible deliberately to allow someone to die in order to
use his body for a medical procedure that would save many lives? It might be
suggested that the distinction between agency and allowing is relevant when what is
allowed to happen is itself aimed at. Yet this is not quite right, because there are
cases in which it does make a difference whether one originates a sequence or only
allows it to continue, although the allowing is with deliberate intent. Thus, for
instance, it may not be permissible to deprive someone of a possession which only
harms him, but it may be reasonable

end p.85

to refuse to get it back for him if it is already slipping from his grasp.

Cf. Philippa Foot, 'Killing, Letting Die, and Euthanasia: A reply to Holly Smith Goldman', Analysis,

41/3 (June

1981

And it is arguable that non-voluntary passive euthanasia is sometimes justifiable
although non-voluntary active euthanasia is not. What these examples have in
common is that harm is not in question, which suggests that the "direct", i.e.,
deliberate, intention of evil is what makes it morally objectionable to allow the beggar
to die. When this element is present it is impossible to justify an action by indicating
that no origination of evil is involved. But this special case leaves no doubt about the
relevance of distinguishing between originating an evil and allowing it to occur. It was
never suggested that there will always and everywhere be a difference of
permissibility between the two.
Having defended the moral relevance of the distinction which roughly corresponds to
the contrast between killing and allowing to die, I shall now ask how it affects the
argument between those who oppose and those who support abortion. The answer
seems to be that this entirely depends on how the argument is supposed to go. The
most usual defence of abortion lies in the distinction between the destruction of a
fetus and the destruction of a human person, and neither side in this debate will have
reason to refer to the distinction between being the agent of an evil and allowing it to
come about. But this is not the only defence of abortion which is current at the
present time. In an influential and widely read article, Judith Jarvis Thomson has
suggested an argument for allowing abortion that depends on denying what I have
been at pains to maintain.

Judith Jarvis Thomson, 'A Defence of Abortion', Philosophy and Public Affairs, 1/1 (Fall 1971).

Thomson suggests that abortion can be justified, at least in certain cases, without the
need to deny that the fetus has the moral rights of a human person. For, she says, no
person has an absolute right to the use of another's body, even to save his life, and
so the fetus, whatever its status, has no right to the use of the mother's body. Her
rights override its rights, and justify her in removing it if it seriously encumbers her
life. To persuade us to agree with her she invents an example, which is supposed to

end p.86

give a parallel, in which someone dangerously ill is kept alive by being hooked up to
the body of another person, without that person's consent. It is obvious, she says,
that the person whose body was thus being used would have no obligation to
continue in that situation, suffering immobility or other serious inconvenience, for any
length of time. We should not think of him as a murderer if he detached himself, and
we ought to think of a pregnant woman as having the same right to rid herself of an
unwanted pregnancy.
Thomson's whole case depends on this analogy. It is, however, faulty if what I have
said earlier is correct. According to my thesis, the two cases must be treated quite

differently because one involves the initiation of a fatal sequence and the other the
refusal to save a life. It is true that someone who extricated himself from a situation in
which his body was being used in the way a respirator or a kidney machine is used
could, indeed, be said to kill the other person in detaching himself. But this only
shows, once more, that the use of 'kill'is not important: what matters is that the fatal
sequence resulting in death is not initiated but is rather allowed to take its course.
And although charity or duties of care could have dictated that the help be given, it
seems perfectly reasonable to treat this as a case in which such presumptions are
overridden by other rights—those belonging to the person whose body would be
used. The case of abortion is of course completely different. The fetus is not in
jeopardy because it is in its mother's womb; it is merely dependent on her in the way
children are dependent on their parents for food. An abortion, therefore, originates
the sequence which ends in the death of the fetus, and the destruction comes about
'through the agency' of the mother who seeks the abortion. If the fetus has the moral
status of a human person then her action is, at best, likened to that of killing for spare
parts or in Rescue II; by contrast, the act of someone who refused to let his body be
used to save the life of the sick man in Thomson's story belongs with the scarce drug
decision, or that of Rescue I.
It appears, therefore, that Thomson's argument is not valid, and that we are thrown
back to the old debate about the moral status of the fetus, which stands as the crucial
issue in determining whether abortion is justified.

end p.87

Morality, Action, and

Outcome

Philippa Foot

No decision is more important for practical ethics than that by which we come to
embrace or reject utilitarianism. For although non-utilitarian principles are apparently
deeply embedded in our ordinary morality, theoretical justification often seems hard
to find; and some common intuitions are in danger of being disregarded on
theoretical grounds. I want to consider two of these intuitions, and to defend them.
The first is that there is a morally relevant distinction between what we do and what
we allow to happen, and the second that there is a similarly relevant distinction
between what we aim at and what we foresee as the result of what we do. I believe it
is rather generally thought that the moral relevance of these distinctions is impossible
to maintain. I shall, however, deny this, arguing that both differences are defensibly
as well as widely recognized in the moral judgements we ordinarily make.
Let us consider first of all the distinction between 'doing' and 'allowing', the moral
judgements which seem to depend upon it, and the nature of the distinction itself. It is
implied, it seems, in many decisions about what it is right to do, e.g. in cases dealt
with in medical ethics. So, for instance, if some medical resource is in short supply
and it would be possible

end p.88

to deploy it either to save a number of patients or to save one, then the policy would
be to save as many as possible. It does not follow, however, that any and every
decision could rightly be taken which resulted in the same net saving of lives. For
although it might be called a regrettable moral necessity that a smaller group should
be left to die while a larger group was saved, this could not be said in the case where
the few were to be killed. It is not respectably believed that medical experimentation
is justifiable so long as the benefit to some outweighs the cost to others; nor do those

doctors whose patients will die without organ transplants think that they should
secretly kill one healthy person if his body could save several. And while it might be
right to leave one injured person to die by the road if hurrying to the rescue to
several, it would be outrageous to drive over a recumbent person in order to reach
them in time. It seems, therefore, that the 'negative duty' of non-interference is
stronger than the 'positive duty' of assistance.

I have discussed this idea more fully in 'The Problem of Abortion and the Doctrine of the Double

Effect', Oxford Review, 5 (

1967

); repr. in Philippa Foot, Virtues and Vices (Blackwell,

1978

The same distinction between 'doing' and 'allowing' appears again when we think of
what someone is obligated to sacrifice, on the one hand to avoid being the agent of
some mischief to another and on the other to prevent or remedy the same evil when
it is in no way due to him. We are inclined to think that no one is required to bring
financial ruin on himself in order to give funds to Oxfam; but to export poisonous food
even to stave off ruin is clearly another thing.
So much is implied in our ordinary moral judgements; but it is not easy to see just
what principle is at work. Is the operative difference the difference between act and
omission, between what someone does and what he refrains from doing? Obviously
this cannot be it. For then it would be possible to change the moral character of
certain trains of events by such simple expedients as building respirators which
needed to be turned on each day. The difference between act and omission is in fact
irrelevant to any moral issue except in so far as it corresponds to the distinction
between allowing something to happen and being the agent to whom the happening
can be ascribed. The difference we should have in

end p.89

mind is roughly this: in the case of allowing, a train of events must already have
started or be on the horizon; an agent who could stop or prevent it does not do so,
and therefore allows it to go on. In the other case it is he who initiates a sequence, as
for instance by shooting someone who dies of the shotgun wound, or by pushing into
the water someone who then drowns.

In some cases the agent will not start a sequence but rather keep it going when it would otherwise

have come to an end. Then he initiates a new stage of the sequence rather than the sequence itself.

No one who has worked on these questions in vexed areas such as medical ethics
can think that the distinction between allowing something to happen and being its
agent is simple or clear-cut, if only because what goes on in the world is not neatly
divided into differentiated sequences.

One interesting complication is that diverting a harmful sequence from one victim to another seems

not to be viewed in the same way as starting one up. A pilot whose plane is going to crash should
steer from a more to a less inhabited area, and the principle apparently holds even in case it should be
certain that someone will die. It does not follow, however, that one could, as it were, start a flood to
stop a fire.

Nevertheless, it is from the moral point of view a different thing to start up a new train
of events in someone else's life and to refuse to intervene to stop one. This may
strike us again if we think about objections to paternalistic action. For while it is not
normally permissible to seize an object from a man because he will harm himself with
it, it does not follow that there is an obligation to help him to keep such a thing if it is
falling out of his reach.

Cf. Foot, 'Killing, Letting Die, and Euthanasia: A Reply to Holly Smith Goldman', Analysis, 41/3 (June

1981

It seems, then, that the moral character of an action is on occasion affected by the
position of the agent in the causal nexus: by the fact that he is on the one hand the
initiating agent of a sequence or happening, or by contrast merely one who does not

intervene. But whether someone's role in a train of events is permissive or initiatory is
something that makes no difference to its outcome. The good or harm for which the
action is a necessary or sufficient, or necessary and sufficient, condition may be the
same in either case. And therefore utilitarians, who place the whole moral
significance of an action in its production of good or harm, must treat the difference
between initiating and allowing as having no independent influence on morality.

end p.90

So much, for the moment, about the first of the principles mentioned in the opening
paragraph of the present piece as being inconsistent with utilitarianism. Let us now
consider the second; namely the one that allows a moral distinction between what
the agent 'directly intends', that is what he aims at either as means or end, and what
he 'indirectly intends' in foreseeing it as a consequence of his action. The moral
relevance of this distinction has often been challenged.

See e.g. H. L. A. Hart, 'Intention and Punishment', Oxford Review, 4 (Hilary

1967

); repr. in Hart,

Punishment and Responsibility

(Oxford, 1968).

I think, however, that it must be allowed. To be sure it often makes no difference to
the injustice of an action whether an injury which it causes is something the agent
aims at or is something he foresees but has not made the object of his will. A
merchant who sold food he knew to be poisonous in order to make money would be
morally no better than an unemployed gravedigger who deliberately killed to get
trade. Nevertheless, there are circumstances in which it is morally permissible to
bring something about without aiming at it although it would not be morally
permissible to aim at it; even though the balance of benefit and harm in the
consequences remained the same. That this is so is proved, I think, by some facts
about the permissibility of allowing an evil to come on some for the sake of saving
others. For sometimes this is a regrettable moral necessity, as in our previous
examples having to do with scarce medical resources and with the person lying
injured by the roadside. But it does not follow that it would be morally unobjectionable
deliberately to leave someone unattended because his death would allow us to save
others. We said earlier that it would be objectionable to kill even for such a good
purpose, and now we must add that it would also be wrong to serve that same
purpose by deliberately allowing someone to die.

In 'The Problem of Abortion and the Doctrine of the Double Effect' I argued (wrongly as I now think)

that the distinction between direct and indirect intention was irrelevant to moral judgement. Not
surprisingly I was then in difficulties about the wrongness of deliberately allowing a beggar to die in
order to use his body for medical research.

There is, moreover, a very interesting and significant extension to the principle that
bans direct intention in certain cases where indirect intention would be permissible.
For it seems that what matters morally is not only how someone acts, which is what
we have so far been considering,

end p.91

but also how his will is disposed even when this cannot affect the course of events.
To see this one has only to think of a case in which someone is a spectator rather
than a participant in a complex of circumstances where several lives might be saved
through one man's death. For then one sees that while it is wrong to aim at the death
of one in order to save others, and wrong of course to go along with the doing of this
action by somebody else, it is also morally objectionable to be glad, as a spectator, if
it is done; or indeed if such a thing comes to pass without human agency. No doubt
some will deny this conclusion about the mere spectator's wishes; but simpler
examples from more familiar surroundings might make them think again. For would it
not be objectionable if, in a hospital where patients were dying for lack of a

transplant, the death of someone in the hospital with several transplantable organs
was automatically treated as good news?
It follows from what has been said in the last two paragraphs that the recognition of a
moral distinction between allowing and initiating evil is not able by itself to take care
of all our non-utilitarian moral intuitions. But would the other principle, the one telling
us that indirect intention may be permissible where direct intention is not, be enough
on its own to do all the work that needs to be done? It is perhaps tempting to think
so, but would nevertheless be wrong. For although many cases of the initiation of
harm are also cases of direct intention the example of the wicked merchants shows
that it is not always so. Moreover, when the evil is merely foreseen and so not
directly intended the morality of the action of which it is an effect may vary according
to whether the agent is or is not the originator of that effect. So, although we may
have to allow an individual to die because we must use the scarce medical resources
for others it would nevertheless be impermissible to manufacture a curative
substance for the same number at the cost of sending fumes to poison one. It seems,
therefore, that we need both the principles discussed in preceding paragraphs, and
may of course need others as well, in order to bring moral theory into line with our
ordinary moral thoughts.
There is, I think, no way of amending utilitarianism which will cure the discrepancies
we have noticed. They will remain so long as the moral character of an action,
motive, or any other subject of moral

end p.92

judgement is taken wholly to depend on its causal relation to the general welfare:
whether welfare is conceived as pleasure, happiness, or preferential choice; whether
maximum welfare is taken as an average or a sum total; and whether it is supposed
to be distributed more or less equally. It is another question, however, whether the
trouble is endemic to welfare utilitarianism in particular or whether it rather belongs to
consequentialist theories as a whole. Might the situation be remedied if some other
form of consequentialism replaced utilitarianism? What if, for instance, the violation of
rights should be recognized as one element in the goodness or badness of states of
affairs; the general idea being that the moral character of an action should depend,
as in utilitarian philosophies, on their outcome—on the goodness or badness of
states of affairs in which they play a causal, or a constitutive, role—but that the
judgement of these states of affairs should no longer be directed solely to welfare?
By this amendment it would become possible to count any violation of rights as itself
contributing an element of badness to the total outcome of an action, and in principle
in this 'goal rights system' such badness could be supposed to outweigh any good
the action might produce. Such a provision would take care of many of our
recalcitrant examples because they do in fact have to do with the violation of rights.

Amartya Sen discusses goal rights systems in his excellent article 'Rights and Agency', Philosophy

and Public Affairs

, 11/1 (1981). It would also, I suppose, be possible to recognize other non-welfarist

values within a theory of this general type, saying for instance that there is more badness in the direct
than in the indirect intention of harm.

This interesting attempt to remain within the boundaries of outcome morality while
throwing off the shackles of welfare utilitarianism has the great advantage of giving
rights a more basic role than they can have within the latter theory. The theory fails,
however, to accord with common morality in the verdict it implies in a special range of
cases having to do with the causal interaction of one agent with another. Suppose,
for instance, that you threaten (reliably) to kill or torture several people if I do not kill
or torture one. Then it may be that by doing one of these actions I can lessen the

total number done; and it would follow in such a theory that whatever weight of
badness is given to this kind of action it

end p.93

could be my duty to do it.

Samuel Scheffler has an interesting discussion of this kind of case in The Rejection of

Consequentialism

(Oxford,

1982

), esp. ch. iv.

We can therefore see that no amendment of utilitarianism which retains its
consequentialism can get away from the following implication:that there is nothing so
bad that it cannot be done to prevent others from doing more things of the same kind.
Consequentialist systems have this implication because in basing moral judgement
solely on the evaluation of states of affairs they allow of no distinction between what
an agent does himself and what he allows others to do. Each contributes its quota of
goodness or badness to the total, along with other elements such as happiness or
pain.
So within a consequentialist system even of the goal rights kind a violation of rights
such as murdering or torturing will be justifiable if the alternative is more murdering or
torturing; and it would follow that any powerful and ruthless person could make it
right, could even make it our duty, to do these things, by threatening to do more of
them himself. Some people may think that it would indeed be our duty to give in to
him if enough were at stake; but if they are consistent consequentialists they must
believe the more obviously bizarre proposition that any net balance of 'goodness'
over 'badness' in the total outcome would do.
So far as I know the only way to meet this objection from common morality while
staying within anything that could possibly be called a consequentialist system is to
introduce 'agent-relativity' into the evaluation of states of affairs. By this expedient
each agent would be enabled to value his own performance of an action differently
from its performance by others; so that each could think it a worse total outcome
when he killed or tortured than when someone else did, other elements of the
situation always remaining the same. Amartya Sen has suggested this as a viable
account of the matter.

'Rights and Agency', sects. vi and vii.

But it seems to me that it is implausible. For it is not true that one thinks that a given
act is worse when done by oneself than when done by another, unless of course
there is some relevant difference between us, as when only one of us is a doctor or a
parent or a friend. It is after all generally agreed that an action done by one individual
is not, except in such cases, morally different from the

end p.94

same action as done by someone else. I do not refuse to kill or torture to prevent
others from killing or torturing because I think that killing or torturing is, in the ordinary
sense, worse when I do it than when they do. So what is it that one is supposed to
think about one's own action as opposed to that of others? Sen speaks of 'evaluating'
differently the states of affairs of which each is a part, suggesting as an illustration
that if Othello and another gave a moral evaluation of the state of affairs in which
Othello had murdered Desdemona the murderer himself would think it morally
worse.

Ibid. 29-30.

And he also uses the idea of an agent-relative moral judgement to explain, e.g., why
someone with moral objections to a certain kind of work might refuse to do it himself
though he felt no obligation to influence others similarly placed. The agent is

supposed to think that it is worse if he takes the job than if someone else does, even
when their situations do not relevantly differ.

Ibid. 24-6 and 32.

It is to be morally worse from his point of view. But it is hard to know what this
means. Sen suggests a comparison with an aesthetic proposition to the effect that an
object is beautiful (he must surely mean 'looks beautiful') from a certain position.

Ibid. 35-6.

But the comparison will not go through. Anything which is, visually, beautiful is so
because it looks beautiful. This implies that it looks beautiful from some distances
and from some angles but not that it must look beautiful from all; and the question
therefore arises of whether it looks beautiful (or of how beautiful it looks) from some
particular place. But it would be impossible even to start to give a similar explanation
of moral badness from an agent's point of view because there is nothing that stands
to being morally right or wrong as looking beautiful from certain positions stands to
being beautiful. I believe therefore that Sen has failed to explain a concept which is
central to his attempt to reconcile consequentialism with our moral intuitions.
It seems, then, to go back to the main line of the argument, that it is impossible to
believe in utilitarianism or any other form of consequentialism and at the same time
to hold on to more or less well-entrenched moral opinions. And this naturally raises
the question: why believe in those theories? Why do they seem to be plausible? Why
in particular does

end p.95

utilitarianism seem to be so in spite of the difficulties it creates for practical morality?
Different people no doubt have their own reasons for believing in utilitarianism, and
some think that they have proof of its truth. But many more appear to embrace the
theory because they are convinced that utilitarian morality is the only rational
morality, and we should ask about the cause of this conviction. It lies, I think, first of
all in the fact that utilitarians insist that moral goodness must somehow be connected
with what is good to and for human beings. But it also lies in the utilitarian, or
basically consequentialist, idea that it must always be right for an agent to bring
about the best state of affairs that is within his reach. And it is this last thought that
makes the theory discordant with our ordinary moral intuitions. For the question
raised is why any action should not be done if the doing of it will produce the best
state of affairs or total outcome; and it is exactly this which tends to make us think
that it must be morally permissible, or even obligatory, to do to individuals anything
that 'needs to be done for the general good'. How can it be right, we ask ourselves, to
choose to produce a state of affairs less good or worse than another that is equally
within our reach? If this state of affairs is the best state of affairs then we ought to
produce it. To be sure we are not always able to do what would bring about what is in
other contexts

referred to as the best state of affairs, but that is because there may

be restraints on our actions coming, e.g., from law, morality, or etiquette. When it is a
question of bringing about the 'morally best state of affairs' there can be no such
restrictions because only moral restraints could operate against morality, and only an
irrational moral rule could stand against the production of the morally best state of
affairs.
It is such thoughts that tend to make utilitarianism seem uniquely rational and
therefore irresistible. Those who hold out against it usually argue that it is not always
right to produce, or try to produce, the best state of affairs; but naturally find it hard to
see how indeed it can be right to prefer a worse over a better state of affairs. What
does not occur to them is that with the posing of the question, which they thought to

be on neutral territory, they are already deep within utilitarianism itself. This is what I
shall now try to show.
Let me begin this part of the discussion by making some observations

end p.96

about the peculiar and problematic character of judgements about good states of
affairs when construed as they are apt to be construed by both sides in this debate.
For of what form are they supposed to be? Clearly 'state of affairs' is not a description
like 'game' or 'holiday' or 'committee meeting' giving the possibility of what Peter
Geach has called an attributive judgement of goodness of the ordinary 'good F',
'good G' form.

Peter Geach, 'Good and Evil', Analysis, 17/2 (Dec. 1956).

But nor, again, does the description 'a good state of affairs' refer to a good like
pleasure, or health, or affection, or understanding. So where do propositions about
good states of affairs belong in the language? And in particular how do they get the
connection which they are supposed to have with human choice? In the case of the
two other types of judgements the relationship, although complex, is usually in
principle fairly clear. Good Fs and good Gs may in certain cases (such as good roots
and good claws) have no particular connection with the human will; but where a
connection does exist it is not magical, nor merely linguistic, but depends upon the
way in which those in some standard position are interested in choosing that type of
person, role, or thing. And for the case of a good the conceptual connection with
choice or reasons for choice though hard to explain, and still very much of a
philosophical problem, obviously exists. For the reason for calling pleasure and
health and affection and understanding goods is clearly something to do with the fact
that they are goods to or for people; and thus give these people a reason to seek
them out.
How is it then with good states of affairs? We often speak of them in everyday life—
such locutions are not the invention of philosophers—and nothing then seems clearer
than the fact that we have reason to try to produce the states of affairs we call good
and avoid the ones we call bad. But is not the explanation of this that in these pieces
of everyday, non-philosophical, usage we use expressions such as 'a good state of
affairs' and 'a good thing' as a way of saying how things fit in with our interests and
with things we are interested in? And of course this makes those particular usages
quite unsuitable as a pattern for the way in which the expressions might function in
moral contexts, where a more

end p.97

impersonal reference would be required.

I have discussed this topic at more length in 'Utilitarianism and the Virtues', Presidential Address to

the American Philosophical Association, Pacific Division, Proceedings and Addresses of the American
Philosophical Association

, 57/2 (Nov.

1983

). See also ch.

of this volume.

I do not know why these impersonal uses of 'good state of affairs' and likewise 'good
thing' and 'It is good that . . . ' are supposed to be unproblematic. It seems to me that
there is every reason for being suspicious of them as they are talked about in
philosophy, where it is supposed that judgements of this form, though having no
original connection with anything that a particular agent sees as a good, or with any
object of his interest or desire, are nevertheless held to compel, persuade, or at least
give him reason for choice.
Let us consider, then, what use there can be in moral contexts for expressions such
as 'a good thing' and 'a good state of affairs'. There seems to be no doubt that kindly
people agreeing with each other in their attitudes will take for granted that certain
occurrences—such as natural disasters for instance—produce 'a bad state of affairs';

and among those who accept morality a certain modicum of kindness will be
assumed. It will also be taken for granted that the avoidance of suffering is among
the ends which can determine what it is right to do. This must be so given any
morality in which benevolence is recognized as a virtue; and a further end is implied
by the acceptance of a morality in which the virtue of justice is held to include fighting
against injustice as well as performing just acts. We find these ends, and perhaps
others, deeply embedded in our morality, and for this reason there is a special way of
talking about states of affairs as good and bad, particularly when a moral issue is
actually being discussed. In such contexts, unlike those of which we spoke earlier,
there is implicit reference to ends belonging to moral persons as moral persons,
rather than to idiosyncratic or personal aims.
It is not, therefore, to be denied that there is a special way of talking about good and
bad states of affairs in what we might broadly call moral contexts. But this in itself
does not mean that we are back with the old puzzle about how it can be right not to
do anything and everything which will be for most people's good. For it is one thing to
think that morality dictates an end such as the happiness of others, and even directs
us, on

end p.98

occasions, towards doing as much good as we can, and quite another to think that
any action is morally permissible if directed to this end. For what we have said so far
does not give any reason why the pursuit of the end should not be restricted by moral
rules forbidding, e.g., certain kinds of interference with individuals even for the sake
of the general good. The fact that benevolence is a virtue giving persons who
possess the virtue certain ends does indeed determine a special use of expressions
such as 'a good state of affairs' in moral contexts. (And as already mentioned the
virtue of justice can also play its part.) But there is nothing in the virtue of
benevolence licensing us to say that it is 'a good thing' when a benefit comes to
many through injustice to a few. So benevolence gives us no reason to say, for
instance, that it would be a 'good state of affairs' or 'good total outcome' if the
sacrificing of a few experimental subjects allowed us to get cancer under control. The
operation of benevolence is circumscribed by justice, and even the end which the
virtue prescribes is qualified, in that we are not told to be glad that good should come
to some when it comes through evil to others.
The fact that benevolence is a virtue, and a virtue which dictates attachment to the
good of others, does not, then, give morality a universal end or goal; and the same is
true of other parts of morality which have to do with what must be aimed at or
desired. Thus the moral requirement to fight injustice does not imply that one must,
or may, fight it by any means. And if there is a separate requirement having to do
with the promotion of good action in others (which perhaps there is), this too can be
circumscribed. So if, at this point, we return to consider the morality of doing some
apparently bad action such as torturing, or killing innocent people, to stop someone
else doing more acts of the same kind, and think that it must be right to do it, we can
see where we have gone wrong. For whether we think of the moral objection to such
actions in terms of suffering inflicted, injustice, or immorality, we have no reason to
suppose that our sole moral duty is to minimize it. We are not even mandated to
minimize immorality. For moral injunctions can be like orders which say not 'See to it
that there is less shouting' (which might be obeyed by shouting) but rather 'don't
shout'. There is thus nothing puzzling about the fact that many moral directives give
rules for an agent's

end p.99

own conduct and tell him nothing about what he is to do about influencing the actions
of others. In so far as one is required to see that other people behave well these
requirements come from other parts of morality and have no priority over moral rules.
The argument of the last few paragraphs has suggested that the use of the notion of
a good state of affairs is very limited in moral contexts. This restriction may, however,
be disputed, on the ground that any reasonable recognition of rules in morality must
allow that they may sometimes be broken, and must therefore deploy the notion of
better and worse states of affairs in deciding when that may be. This has recently
been argued by Amartya Sen, who insists that any theory which allows deontological
restraints to be overridden in certain circumstances must collapse into a form of
consequentialism, because if it is determined that a constraint may be overridden in
specified circumstances (as, e.g., a right violated when it will save lives) then the
badness of the violation is being weighed against that of non-violation.
For example, it can be specified that if the badness of the state of affairs resulting
from obeying the constraint exceeds some 'threshold', then the constraint may be
overridden. Such a threshold-based 'constraint' system must rest ultimately on
consequential analysis, comparing one set of consequences (badness resulting from
obeying the constraint) with another (badness of violating the constraint itself, given
by the threshold), and its distinguishing feature will be the particular form of the
consequence-evaluation function.

'Rights and Agency', 6-7 n. 8.

This is an interesting argument, but I think it is mistaken. To meet Sen on his own
ground I shall assume that moral rules can indeed be broken on occasions, which is
certainly true of some even if not of all. So why do we say that in these
circumstances this rule but not that one could be broken, or that this rule could be
broken in these circumstances but not in those? Sen supposes that we make the
decision by estimating the goodness and badness in the total outcomes of the
alternative choices. But this seems to me to be either vacuous or wrong. If it merely
says that our decision may be put in terms of 'better' and 'worse'('It is better to do
without a cure for cancer than to perform this experiment') it is vacuous. But if it says
something about the way in which the decision is arrived at it is

end p.100

dubious. For what entitles Sen to assume that this is how we judge rightness, i.e. by
comparing total outcomes; when what is, or should be, at issue is the concept of a
good total outcome

as intended here?

It seems, then, that we can accept the idea that we talk in special ways in moral
contexts about good and bad states of affairs without implying anything that could
make the existence of non-utilitarian principles problematic. Some virtues do indeed
give us aims, but nothing from within morality suggests the kind of good state of
affairs

which it would seem always to be our duty to promote. And why indeed should

there be any such thing?
It will be helpful here to consider a parallel between the case of morality and the case
of etiquette. Why is it, we may ask, that we do not have good and bad states of
affairs from the point of view of etiquette? The notion seems incomprehensible. But
why is it so? It is so because a good state of affairs from the point of view of etiquette
would be one which from the point of view of etiquette it must be right to aim at or
produce. But in fact there can be no state of affairs which stands in this position
because even if there are aims prescribed by etiquette (as, e.g., avoiding causing
social embarrassment), etiquette is also a matter of following rules, and the rules
circumscribe the manner in which the aims may be followed. So what of morality? A

morally good state of affairs would, it seems, be a state of affairs which morally right
or good action must aim at or produce. But since what is morally good or right is what
a good moral system calls good or right there can be a state of affairs which stands
in this position only if a good moral system contains no radically non-utilitarian rules.
So if a good moral system does indeed contain rules of the kind we have been
discussing then there cannot be a good state of affairs in the sense which
consequentialism requires. There is, as we saw, a limited use for the concept within
morality; but a virtue such as benevolence does not give an end to which all moral
action must aspire.
So far, then, the argument has tended to resist the encroachment of any form of
consequentialism on the 'mixed' aim-and-rule morality that we actually seem to have.
But nothing has yet been said of the rationale of this partly deontological system. If
we do in fact recognize virtues such as justice which consist mainly in adherence to
rules of conduct, as well as those like benevolence which we might call virtues of
attachment, why

end p.101

do we do so? Will utilitarianism not re-establish itself when it comes to defending the
morality we have?
This is an important line of thought: it is responsible for the faith of many utilitarians,
and the answer to it goes to the heart of the debate. So let us see why it is that a
non-utilitarian morality might be thought to need a utilitarian defence. The idea is, I
think, that a morality can be shown to be a good morality only by being shown to do
better

than its rivals; that is to work better or produce a better state of affairs. And this

generates the old problem about individual actions which do not, because of
exceptional circumstances, have the usual effect. For then violation (perhaps secret
violation) of the code could produce a better state of affairs than obedience, and it
will seem irrational to insist on adherence nevertheless.
The crucial assumption in this argument is the one that links the goodness of a moral
system to a prior goodness in states of affairs. For states of affairs are things that can
be brought about or influenced by actions, and this is why we have only to suppose
strange causal circumstances to get a violating action producing 'the best state of
affairs'. The question we should ask is, therefore, why this criterion of goodness
should be assumed for a moral system. Perhaps some think that there just are good
states of affairs and that this is a fact which can hardly be ignored in moral
judgement, but I hope that the argument in the early part of this paper has helped to
make this seem more problematic. Others believe that the moral point of view relates
to the unique purpose of maximizing welfare, and try to prove it, but it is significant
that the general line of argument outlined above seems irresistible to many who are
not actually satisfied by any such proof, and therefore have no warrant such as Hare
and Harsanyi believe themselves to have for operating with the idea of states of
affairs which are good from the moral point of view.

See R. M. Hare, Moral Thinking (Oxford, 1981) and John C. Harsanyi, 'Morality and the Theory of

Rational Behaviour', Social Research, 44/4 (Winter 1977); repr. in Amartya Sen and Bernard Williams
(eds.), Utilitarianism and Beyond (Cambridge, 1982). Hare's position has been criticized, to my mind
effectively, by Thomas Nagel, London Review of Books (1-15 July 1982).

But is it then so obvious that the criterion of goodness in a moral

end p.102

system must lie in its relation to a state of affairs that is judged to be good? Perhaps
it will seem obvious so long as no alternative comes to mind. But of course there are
alternatives which make a clean break with the idea of morally criterial 'good states of
affairs'.

To see that this is true we have only to look at the way in which Rawls, for instance,
sees the distinction between a good moral system and a bad one. For in his moral
philosophy any good moral system is necessarily a fair one, and a fair system one
which operates on principles which would be chosen by rational agents in the original
position. Thus a good moral system will be one that has a certain characteristic which
is not a causal property, and there need be no place in this theory for an original
judgement about good states of affairs.

John Rawls, A Theory of Justice (Cambridge, Mass., 1971).

The same is true of Scanlon's more thoroughgoing contractualism, under which a
moral rule's justification depends on its acceptability (a rather complex kind of
acceptability) by, or on behalf of, every individual whatever his position in the
world.

T. M. Scanlon, 'Contractualism and Utilitarianism', in Sen and Williams (eds.), Utilitarianism and

Beyond

. I find this a bold and very suggestive piece.

It is not necessary to commit oneself to either system to see in outline how a
justification might be produced for the two principles of nonutilitarian morality
discussed earlier in the present paper. Suppose for instance that we simply start from
the rough idea that a good moral system must be one that could use a demand for
reciprocity to urge conformity on any individual; and that the necessity that it produce
benefit to him should be linked to this condition if to no other.

Cf. ibid. 118-19.

Then it would be intelligible that the more a morality rendered benefits from which
each and every person stood to gain, the more acceptable, and so far forth the
better, the system would be. And of course there are such benefits from the
existence of a morality which refuses to sanction the automatic sacrifice of the one
for the good of many because it secures to each individual a kind of moral space, a
space which others are not allowed to invade. Nor is it impossible to see the rationale
of the principle that one man should not want evil, serious evil, to come on another
even to spare

end p.103

more people the same loss; it seems to define a kind of solidarity between human
beings, as if there is some sense in which no one is totally to come out against one of
his fellow men.

Perhaps it is this idea that is partly responsible for the peculiar outrage that we feel about torture.

The principle must obviously be qualified to take care of the case of guilty men. Yet witnesses to
judicial executions have reported a primitive reaction of horror at the fact that no one went to help the
man in the electric chair, as if it was impossible really to believe that the assembled company wanted
him to die.

In both cases the good of the rule is a good that comes from having the system. But
the justification is not, as with rules that limit the direct pursuit of the general good in
rule utilitarian systems, that those who accept them will be most likely actually to
bring about the best result. For if this were it the supposition of exceptional
circumstances would give cases in which the agent would be obliged to break the
rule.
What I am suggesting is that the concept of 'the best state of affairs' should
disappear from moral theory, though not, as explained earlier, from all talk in moral
contexts. If this would allow some of the insights of rule utilitarians to be validated, so
much the better. But the starting-point would be so different that the results would be
unlikely to be the same. It has been suggested here that one criterion for a good
moral system is that it should be possible to demand reciprocity from every individual
because of the good the system renders to him. But I am sure that this is not the only

condition for a good moral system. It has also, for instance, to be such that anyone
can conform to it and still live well in the ordinary, non-moral, sense.

Nietzsche's suggestion that some men's lives are necessarily deformed by obedience to morality is

not, I think, one that we could see as leaving morality intact.

This condition may well be what limits the demands of altruistic action, and a whole
new non-utilitarian enquiry should open up here.

end p.104

Von Wright on

Virtue

Philippa Foot

Earlier it had been rare to find a discussion of the virtues in a contemporary work of
moral philosophy in the analytic tradition, and it was with pleasure that, in 1963, one
discovered a whole chapter in von Wright's Varieties of Goodness devoted to this
topic. The study of ethics had been, for philosophers such as Aristotle and Aquinas,
largely a study of the virtues, and as the foundations of Western morality had been
laid down by Greek and Christian thinkers it would be surprising if we could neglect
this topic without loss. It was not, however, in any revivalist spirit that von Wright took
us back to the virtues. One might think that by neglecting this study we would have
managed to lose an important body of knowledge. Von Wright, however, suggests
that we had rather allowed a backward part of philosophy to remain in an
underdeveloped state. Kant's dictum about logic—that it had made no real progress
since Aristotle—could, he says, be applied with at least equally good justification to
the ethics of virtue, and he seems to see the future development of the philosophy of
the virtues in terms of radical change. So he sets out to shape a new concept of
virtue

and one sees how far von Wright is prepared to go in throwing over old

doctrines when one realizes that he is happy with a definition which excludes two of
the four cardinal virtues of ancient and medieval morality.
Why is von Wright so dissatisfied with the traditional theory of the virtues? Perhaps
this is not made altogether clear in his discussion, but

end p.105

several themes appear. In the first place, von Wright himself seems to be somewhat
preoccupied with the need to distinguish a virtue, in the sense in which we now use
that term, from an art or a skill, and he believes that Aristotle, 'misled by the
peculiarities of the Greek language', did not see how different they are. We might
start by considering what von Wright has to say on this subject, and thinking about
the justice of his charge.
Von Wright's own answer to the question 'How does a virtue differ from an art or a
skill?'is as follows. If one possesses a skill, or is master of an art, one has what he
calls 'technical goodness'and technical goodness is a matter of being good at
performing some specific activity, such as running, skiing, or singing. A virtue must
be different because there is no specific activity connected with any virtue, and
therefore nothing for a man of virtue to be good at.
The lack of an essential tie between a specific virtue and a specific activity
distinguishes virtue from that which we have called technical goodness. We attribute
technical goodness or excellence to a man on the ground that he is good at some
activity. But there is no specific activity at which, say, the courageous man must be
good—as the skilled chess-player must be good at playing chess and the skillful
teacher must be good at teaching.

The Varieties of Goodness (New York: Humanities Press, 1963), 139.

Is von Wright correct in the account he gives of the possession of a skill or art?
Clearly he is right in saying that it is a matter of being good at doing something,
though one might query whether what the skilful man is good at is always an activity
in von Wright's special sense, and we should pause to consider the distinction
between activities and acts which he often refers to in the discussion of the virtues.
What von Wright calls an act is the intentional production of a certain result, as, e.g.,
lighting a cigarette or opening a window.'Acts are named after that which I have
called the results of action, i.e. states of affairs brought about or produced by the
agent in performing the acts.'

Ibid. 141.

An activity, on the other hand, is related not to an event but to a process. 'Events
happen, processes go on. Acts effect the happening of events, activities keep
processes going.'

Norm and Action: A Logical Enquiry (London: Routledge & Kegan Paul, 1963), 41.

Acts, says von Wright necessarily leave an imprint on the world whereas activities
may or may not do so.

end p.106

Activity is not internally related to changes and to states of affairs in the same
manner in which acts are related to their results. Activity, however, may be externally
or causally related to changes and states which are consequences of performing this
activity. Running need not leave any 'imprint' on the world, but smoking may leave
smoke.

Ibid.

It is at first sight puzzling that von Wright, who has made this distinction between act
and activity, says that every skill or art has to do with being good at some activity. It
certainly is so in the case, e.g., of the singer, who must do well in the activity of
singing rather than produce some result in the world. But what about a doctor's skill,
which seems to lie in his ability to bring about a cure in the patient? Why should we
not say that what he is good at is performing the act 'curing a patient' which is
defined in terms of this result? He is able to do this difficult thing, and in
circumstances where others could not, though of course the skilled doctor need not
always produce the result, even in cases within his general competence; the
expression 'being good at effecting a cure' allows for some failures, particularly where
external factors intervene, and this gives us no reason to deny that what he is good
at is producing the result.
Perhaps von Wright would reply that there are some activities defined in terms of the
result produced, but which are activities none the less. It would seem to be
necessary to give some such account of the meaning of a form of words such as "He
was writing a letter". In terms of the distinction made by Kenny and others between
performances and activities this names a performance rather than an activity.

See e.g. Anthony Kenny, Action Emotion and Will (London: Routledge & Kegan Paul, 1963).

Yet "He was writing a letter" implies that the result had not yet been accomplished,
and does not imply that it ever would be accomplished, so that one would expect von
Wright to deny that any act was named. If activities and acts divide the field between
them there should, therefore, be activities named after the intended result, and it
could be said that some skills were a matter of being good at the performance of
such activities.
I shall leave this problem unsolved and will accept at least the outline of von Wright's
account of skills and arts. It still remains to be seen

end p.107

whether he has drawn the right distinction between skills and arts on the one hand
and virtues on the other. Is he right that the difference depends on the fact that to a
virtue no specific activity corresponds? Is it indeed true that none of the virtues has
an essential connection with a specific activity?
Von Wright argues to this conclusion by insisting that when we know that a man is
acting courageously, for example, we do not know what he is doing, since the
activities a man can perform in acting courageously can be so very diverse. Is this
really enough to show that we do not know what he is doing? Are there not many
activities which can be performed by doing any one of a number of things, as a man
who is cooking can be stirring, weighing, reading a cookbook, testing the oven, etc.,
and one who is looking after children can be walking, bathing, joking, nursing, and a
thousand other things? Perhaps it is going too far to say that because we know that a
man who is acting courageously is facing the fearful for the sake of some good we
also know what he is doing, but is it certain that we would want to say that there are
no "specific activities" essentially connected with any of the virtues? Does not
benevolence, for instance, have such a connection with such things as helping? And
if one refuses to call helping a "specific activity" what is one's reason for so restricting
the use of this expression? In any ordinary sense we know what a man is doing if we
know that he is helping even though a great deal is left unsaid.
In any case von Wright seems to be mistaken in thinking that the distinction between
arts or skills and virtues depends on the denial that the latter are connected with
specific activities. For however close the connection between certain virtues and
certain activities a man does not possess the virtue by being good at the activity. The
reason for this was indicated quite correctly by Aristotle when he said that in art he
who errs voluntarily is preferable, whereas in the matter of wisdom, justice, etc., it is
the reverse.

Aristotle, Nicomachean Ethics, bk. 6, ch. 5. See also Aquinas, Summa Theologica, 1a 2ae, q. 57, a.

3 and 4 (First Part of the Second Part, question 57, articles 3 and 4).

Thus, to use Aristotle's example, a grammarian who commits a solecism on purpose
does not give any evidence of deficiency in the art of grammar, whereas no one
could rebut a charge of injustice or folly

end p.108

by saying that he chose to act unjustly or foolishly. A man may be good at φing—may
possess the relevant skill or art—whether he chooses to φ well or badly: he may
possess the skill but not choose to exercise it; whereas the virtue gives its use.
It may be that there are other things that can be said about the difference between a
virtue and a skill. Traditionally it was held that virtues have to do with human good in
general while skills do not, and Aquinas, for instance, insisted that it is characteristic
of virtues that they cannot be used to bad ends.

Aquinas, Summa Theologica, 1a 2ae, q. 55, a 4 (First Part of the Second Part, question 55, article

4).

But the latter point is controversial, and we do not need to look further than the
consideration raised in the previous paragraph to show that virtues are different from
arts and skills. A skill is shown by the congruity between a man's intentions and his
performance: if he chooses to φ well and is skilful, then by and large he will φ well.
For a virtue on the other hand this congruity between intention and performance is
not sufficient. And nor is it necessary. For it is rather by his intention that a man is
judged virtuous than by the successful accomplishment of what he intends. And this
last point shows one way in which it is quite right to deny that virtues correspond to
specific activities. A man might have the virtue of benevolence, and be acting out of
benevolence on a particular occasion, without actually being engaged in any such

activity as helping, rescuing, etc.; he might be doing that which he reasonably
believed to be helpful but which actually was not helpful at all—like ringing at the
wrong doorbell to rouse a doctor or nurse. I am not, of course, asserting that
benevolence is simply a matter of what a man intends, since failure might be due to
negligent execution, or culpable ignorance such as ignorance of elementary first aid,
and this might show the lack of the virtue itself. Nevertheless, a man who is
displaying the virtue need not actually be engaged in those activities which have an
essential connection with benevolence, and to this extent von Wright is correct in
what he says.
A similar point can be made about the connection between particular virtues and
particular kinds of acts. The virtue of benevolence cannot be defined without
reference to a particular aim. But it is true that the class

end p.109

of benevolent actions does not coincide with the class of acts having this result. It is
therefore true to say as von Wright does that 'Virtuous acts cannot be characterized
in terms of their results, and therefore virtues not in terms of achievements.'

The Varieties of Goodness, 141.

Moreover, there is more to the definition of benevolence than is given by specifying
the intention. First of all the aim of helping another person must not be merely an
intermediate end; there must be no ulterior motive in the case. And secondly
benevolence is not defined as seeking the good of others on any occasion, and by
whatever means. According to the traditional account of the virtues a man who does
a bad act such as perjuring himself in a law court in order to help someone is not
performing an act of any virtue; and even those who would not accept such a close
connection between virtue and good action would admit that no one shows a lack of
benevolence in refusing to do such a thing. I do not know whether this is what von
Wright has in mind when he says at one point that the path of virtue is never laid
down in advance,

Ibid. 145.

but it certainly is true that the path of the benevolent man is not laid down by a
description such as "aiming at the good of others"; he often has to decide whether he
will be failing in charity if he does not do this.
We are now in a position to consider another of the revisions that von Wright wishes
to make in the traditional doctrine of the virtues: his insistence that virtues are not to
be described as dispositions. The argument has to do with his thesis that acts of
virtue do not correspond to specific activities or act categories, and he says explicitly
that dispositions are what the virtues would be 'if there existed act-categories or
specified activities answering to the virtues'.

10 Ibid. 142.

Dispositions, he says, are "inward"things with "outward"criteria, and he thinks that
virtues fail to meet the second condition. I am not myself clear why von Wright insists
that the manifestations of a disposition must be "outward"in a sense that excludes the
manifestations of virtue. There are, it is true, dispositions such as allergies of which
the signs are involuntary happenings like sneezes, and to which a man's thoughts,
intentions, and motives are irrelevant. But there are other dispositions, which von
Wright allows

end p.110

as dispositions, for which this is not the case. If a man has an irascible disposition he
may, when crossed, go red in the face and put on an involuntary scowl, but we
cannot describe the occasion of these happenings without referring to his
understanding of the situation, and the involuntary manifestations of irascibility would

not be very significant without intentional behaviour such as insult, protest, or refusal
of cooperation. Intentional action is a large part of the manifestation of the
disposition. It seems therefore that one should not put too narrow an interpretation on
the condition that the manifestation of any disposition must be something "outer"; and
by the same token there seems no reason to deny that virtues are dispositions.
By far the most radical break that von Wright makes with the past appears, however,
not in his criticisms of traditional theories but in his idea that a new definition of virtue
is needed, and in the offering of a definition that excludes so many of the traditional
virtues. I am not clear why he thinks that such a definition is useful—whether he
thinks, for instance, that no account could be given of a virtue which would embrace
wisdom, justice, and benevolence, as well as such virtues as courage and
temperance. But what is certain is that he claims for the concept that he is, by his
own account, "shaping", that it fits some of the most obvious examples of virtues. 'We
cannot claim that everything which is commonly and naturally called a virtue falls
under the concept as shaped by us. But, unless I am badly mistaken, some of the
most obvious and uncontroversial examples of virtues do fall under it.' And he adds,
'It is therefore perhaps not vain to hope that our shaping process will contribute to a
better understanding at least of one important aspect of the question, what a virtue
is.'

The Varieties of Goodness, 138-9.

According to von Wright's definition of virtue each individual virtue is said to be a form
of self-mastery. To each virtue there is to correspond a particular passion, as fear
corresponds to courage and the desire for pleasures to temperance, and the virtue is
to consist in the mastery of that passion. One notices that von Wright has already
broken with tradition in insisting on a specific passion corresponding to each virtue.
For

end p.111

traditionally a distinction is made between courage and temperance which are "about
passions" and justice which is "about operations". A man who is going to act justly
must indeed have control over his passions, but there is no one passion against
which he is fortified by this virtue. A man who lacks courage must be overcome by
fear, whereas one who acts unjustly may be motivated in many different ways.
Von Wright thinks of each virtue as a form of self-mastery or self-control, and we
must ask whether any such theory could be satisfactory. But first we must notice that
he gives a rather surprising account of what this self-mastery is. It is, he says, the
ability to prevent passions from interfering not with actions directly but with these as
they are affected by judgements. And the judgements involved are a special class of
judgements, that is judgements about what is harmful or beneficial to the agent
himself or to others. (Von Wright equates these with judgements of right and wrong
via the thought that they tell us what it is right or wrong to do with a view to
someone's good.) A passion is, therefore, seen as disorderly or uncontrolled only in
so far as it affects this kind of judgement. If one is courageous, fear does not interfere
with a clear perception of what would benefit oneself or others.
This is, on the face of it, a curious account of the mastery of the passions, since it
seems that these can affect our behaviour not only by corrupting our judgement, but
also by interfering with our performance of what we see clearly to be the best thing to
do. Von Wright will not, however, admit the second case, at least so far as
judgements of our own good and harm are concerned. Where we are inclined to say
that a man is led by present passions to act against his own foreseen long-term
interest von Wright insists that he was not at that time capable of seeing clearly the

disadvantages in what he does. In fact von Wright offer the time-honoured solution to
the problem of akrasia which consists in denying that we do ever truly see the better
and choose the worse. On the other hand he does not deny that we may see clearly
that what we do is harmful to someone else and do it all the same, and I am not clear
how he will argue that such conduct is never caused by our lack of control over our
passions.
I intend to put such problems aside, for quite apart from the possible

end p.112

difficulties in von Wright's special account of the mastery of the passions, there are
difficulties in any theory which equates virtue with self-control.
In the first place the definition is, as already mentioned, one that must exclude from
the class of virtues such things as honesty, charity, fairness, generosity, and
truthfulness. The reason for this is obvious, since these virtues are all concerned with
the readiness to do certain things, such as paying debts, helping others, or telling the
truth, and for this mastery of the passions is a necessary but not a sufficient
condition. If a man is a ruthless self-seeker he may act dishonestly or unjustly or
unkindly with his passions perfectly under control; he sees clearly what he is doing
and coolly proceeds with his plans. He has control over the passions but lacks the
virtues; therefore if a virtue is said to be a form of self-control, honesty, justice, and
so on turn out not to be virtues after all.
Von Wright does not for a moment deny this strange conclusion. Indeed he says
explicitly that neither charity nor justice possesses the features he takes as
characteristic of a virtue and gives arguments like the ones in the preceding
paragraph to show that considerateness is not a virtue in his sense of that word. The
argument about considerateness is interesting in showing just how things would work
out if von Wright's definition of virtue were accepted. There is, he says, a virtue (a
form of self-mastery) forming part of what we call considerateness. A considerate
man must have learned to control his selfish impulses so far as these might interfere
with his judgements about the harm that will come to others through possible action
of his. If he wants to avoid what will harm them, the man with his "virtue"will be well
placed to know how to do so and, we may add, to follow through a decision that that
is what he should do. It will not, however, follow that he acts considerately since he
may not mind harming them. Von Wright's "virtue", which is something for which we
have, I think, no name, is therefore not the virtue of considerateness, and
considerateness is not a von Wright type virtue, since it is not a form of self-control.
Whether or not this is a fatal objection to von Wright's theory depends on the reasons
he has for shaping his new concept of virtue. What would certainly be fatal would be
the discovery that the new concept did not fit

end p.113

even those virtues for which it was bespoken, as, for instance, courage and
temperance. As courage seems to be von Wright's favourite virtue—the one to which
he refers most often in his discussion—we may usefully ask whether courage is
adequately defined in terms of self-control. There is no doubt that courage involves
mastery of the passion of fear. If a man is courageous fear does not lead him to
overestimate dangers, nor to see happiness in terms of safety as a timid man may
do. Nor does he run away where he thinks it would be better to stand. But it is one
thing to say that courage requires mastery of fear and quite another to say that this is
what courage consists in. For if we know that a man has mastered the passion of
fear we know nothing about his ends, except that he is not a fearful, safety-loving
individual. For all that has yet been shown he may be a foolish braggart, or a villain,
and he may show his "courage" (his mastery over fear) in his foolish or villainous

acts. Von Wright does not seem to see any difficulty in this position since he speaks
of the courage 'which burglars or robbers display'.

The Varieties of Goodness, 153.

But I wonder why he dismisses without discussion the opposite point of view. Neither
Aristotle nor Aquinas would allow that an act of a virtue could be a bad act: a virtue is
something of which, as Aquinas put it, no one can make bad use. The courageous
man faces the fearful where he should.
Many people would, of course, agree with von Wright in thinking that there is no
difficulty about the idea of courage used for foolish or villainous ends. And yet if faced
with clear examples they may feel some doubts about so identifying courage with
daring or boldness. Hardly anyone is ready to say that a man who murdered his wife
to get her out of the way would be acting courageously or doing a courageous act so
long as he was sensible of the risk that he ran. They are more likely to use
expressions such as "an act of courage"where the example is slightly different, and
the evil distant from the action concerned, as when, for instance, a man does
something to save his own life or that of his companions in the course of some
wicked enterprise, such as an unjust war. But this tells against rather than for the
idea that courage is simply mastery of fear; why should the presence of immediate
good ends and the remoteness of

end p.114

the bad ones make any difference if the goodness and badness of action does not
come in to the definition of courage? Perhaps some compromise may be suggested,
by which we shall indeed refuse to call the act of murder a courageous action but will
say of the murderer that he "shows courage". But then one must ask "What is this
courage that he shows?"If it is identical with daring why should a daring act not be
the same as a courageous act? If we say that the disposition of which the act is the
criterial manifestation is indeed boldness or daring rather than courage, but that it is
nevertheless evidence that the man also possesses courage, we shall want to know
how the inductive basis for such a judgement is reached. Have we observed that
men who do bold and daring acts of the kind we have been considering also do acts
of courage? The most that can be said for this argument is as follows. A man who is
to act courageously must have the capacity to face the fearful, and this capacity is
shown by the performance of acts of boldness and daring, however foolish or
villainous. Therefore that capacity which is a necessary condition of courage may be
shown in bad or foolish acts. Someone who does them may be unready to face the
fearful for good ends, but not because he is incapable of mastering his fears.
I believe, then, that there is a closer connection between the concept of courage and
that of good action than von Wright allows. I do not, of course, mean that courageous
actions have to be what people normally think of as morally good actions such as
acts of charity or justice, since it is clear that a man may act courageously (may do a
courageous action) for the sake of some good in his own life, as he might submit to
the amputation of a limb under appalling conditions in order to save his health or his
life. But I think that there is a sense of"good action"in which courageous actions must
be good actions, and a sense which definitely excludes the action of the villain or the
foolish braggart. To face the fearful where one should not does not indeed show lack
of courage, and therefore courage is not to be defined as "getting it right"about when
the fearful is faced. But only good acts are courageous actions, and if the villain or
fool is courageous, as he may be, this is not shown by his bad or silly acts.
In this essay I have concentrated on those parts of von Wright's

end p.115

discussion of the virtues in which he criticizes and departs from the traditional
doctrine of the virtues, and have suggested that there is more to be said for the point
of view of Aristotle and Aquinas than he allows. But I do not mean to suggest that the
subject was brought to finality by these philosophers. Von Wright is clearly correct in
thinking that the topic needs to be reopened and developed, and he himself has said
many true and interesting things that will contribute to this development.

end p.116

Locke, Hume, and Modern Moral

Theory:

A Legacy of Seventeenth- and Eighteenth-Century Philosophies of Mind

Philippa Foot

'Tis one thing to know virtue, and other to conform the will to it. In order, therefore, to
prove, that the measures of right and wrong are eternal laws, obligatory on every
rational mind. . . [w]e must also point out the connexion betwixt the relation and the
will. . .
(David Hume, Treatise of Human Nature

David Hume, A Treatise of Human Nature (1739-40), ed. L. A. Selby-Bigge, 2nd edn., ed. P. H.

Nidditch (Oxford: Clarendon Press, 1978), iii. i. 1, p. 465. All subsequent references to Hume's
Treatise

are to this edition.

Where there are problems about the availability of works quoted in this paper, I have occasionally
given alternative references, enclosed in square brackets.

)
Most philosophers writing today would be surprised to be told that their own
philosophical psychology had anything much to do with that of late seventeenth- and
early eighteenth-century philosophers. No one indeed denies that interest in the
philosophy of this period is still very great and not at all declining, if only because in
metaphysics Locke's theory of primary and secondary qualities, and his theory of real
essences, are widely

end p.117

discussed; and in moral philosophy Hume is a commanding, even dominating, figure,
whose arguments we can never quite satisfactorily refute. The philosophies of mind
of these two great empiricists have, however, come to seem more than a little quaint.
We are no longer interested in the attempt to show the origin of psychological
concepts in simple ideas or impressions presented in experience; and Wittgenstein
has taught us to regard with suspicion the belief, firmly held by both Locke and
Hume, that each person knows only from his own introspection what it is to think, to
feel, or to desire.

See Ludwig Wittgenstein, Philosophical Investigations (1953), trans. G. E. M. Anscombe (New York:

Macmillan, 1968), passim, but esp. sect. 243 ff., and e.g. 316, 412, 551, 580, 587. Also pt. ii, xiii.
In writing of modern philosophy I confine myself throughout to philosophy of the modern analytic
school. It is absolutely necessary to deal with this separately from, e.g., phenomenology.

In this paper I shall argue that although we have, it is true, rejected a great deal of
the philosophical psychology of Locke and Hume, there is a part of it which we still
more or less take for granted. If the debt is little acknowledged, this may be because
it has come to appear to many simply as incontrovertible, uncontentious truth. I am
thinking here of a part of the philosophy of mind which belongs to the theory of
volition and thus to the theory of action; and not surprisingly it is in the related field of
moral theory that its influence appears. I shall describe the views of Locke and Hume
on the subject of the determination of action, showing how these theories affected
their own (very different) moralities. I shall point out an underlying similarity in their

philosophies of action which made it impossible for either of them to give a proper
account of moral motivation, and shall suggest that our own thinking on the subject of
moral judgement would be improved if we made a more radical break with Locke and
Hume.
To begin, then, with this part of the philosophical psychology of John Locke. What is
interesting in this connection is his account of the determination of action by desire;
there are several other parts of the Lockean philosophy of mind and self which are
relevant to modern theories of ethics, as for instance his doctrine of self-identity, or
again his theory of free will; but these I will set aside. Naturally, however, Locke's
theory

end p.118

of the determination of the will to action must be seen in the context of his general
empiricist epistemology. What Locke is out to do, in the theory of action as
elsewhere, is to measure the reach of the understanding and to show that we can
account for all our knowledge in terms of experience and the combination of ideas in
the mind. He insists that all our concepts can be seen as combinations of simple
ideas given to us by sense and by reflection, 'sense' being the operation of our sense
organs, and 'reflection' the reflexive movement that we make in order to observe the
operations of our minds.
Where, we may ask, do mental elements such as desires and emotions belong in this
scheme of things? To Hume they would be impressions of reflection; but the
structure of Locke's theory of mind is rather different. Desires and emotions are, he
says, modifications of the simple ideas of pleasure and pain; and he classifies these
as being not only ideas of reflection but also ideas of sense. What Locke means by
this is that pleasure and pain can arise either from the senses (as when fire warms or
burns us) or alternatively from thoughts that bring joy or grief. He makes his position
quite clear in a sentence from book ii of the Essay concerning Human
Understanding

: 'By Pleasure and Pain, I must be understood to mean of Body or

Mind, as they are commonly distinguished; though in truth, they be only different
Constitutions of the Mind, sometimes occasioned by disorder in the Body, sometimes
by Thoughts of the Mind'.

John Locke, An Essay concerning Human Understanding (1690), ed. Peter H. Nidditch (Oxford:

Clarendon Press, 1979), ii, xx. 2, p. 229. All subsequent references to Locke's Essay are to this
edition. Since Nidditch retains the original spelling, capitalization, and italics of the fourth edition of
1700, quotations from the Essay in the present paper are not uniform in appearance with the extracts
from other writings available only in modernized form. I have chosen to use Nidditch's edition in spite
of this, because it is widely available, and very much the best.

And again, writing of 'Delight and Joy on the one side; and Torment and Sorrow on
the other', he says that 'to speak truly, they are all of the Mind; though some have
their rise in the Mind from Thought, others in the Body from certain modifications of
Motion'.

Locke, Essay, ii. xxi. 41, p. 258.

Such is the framework into which Locke's account of the antecedents of volition has
to fit. He believes that this is how we obtain the simple

end p.119

ideas of pain and pleasure; and it is around pleasure and pain (which he identifies
with happiness and unhappiness) that his theory of action revolves.
Turning now to the considerable complexities of this theory we may be glad to find
one proposition which he unambiguously and consistently endorsed. Locke held the
belief that later came to be known as 'psychological hedonism', according to which all
action is in some way directed towards pleasure or the avoidance of pain. For all his

scorn for Hobbes and his tendency to play down any Hobbesian influence in his own
work, he had in fact taken over Hobbes's belief that human nature is under the
governance of two masters, pleasure and pain. It is true that Locke had freed himself
entirely from Hobbes's materialist account of the origin of these elements of our
experience, giving thoughts and bodily states an equal power to produce them. It is
also true that Locke was much more liberal than Hobbes in his recognition of the
diversity of objects in which men might take pleasure.

In a short undated paper, 'Thus I Think', Locke had given, in a catalogue of lasting pleasures: health,

reputation, knowledge, and 'doing good'. Lord King, The Life and Letters of John Locke (1829), 2nd
edn. (London, 1830), ii. 120; 3rd end. (London: Bohn's Standard Library, 1858), 306-7.

Nevertheless, he says explicitly, and repeatedly, in the Essay both that uneasiness
and satisfaction determine the will and that good and evil 'are nothing but Pleasure or
Pain, or that which occasions, or procures Pleasure or Pain to us'.

Locke, Essay, ii. xxviii. 5, p. 351.

In the Journal for 1676 he writes
The Mind finding in itself the ideas of several objects which, if enjoyed, would
produce pleasure, i.e. the ideas of the several things it loves, contemplating the
satisfaction which would arise to itself in the actual enjoyment or application of some
one of those things it loves and the possibility or feasibleness of the present
enjoyment, or doing something toward the procuring the enjoyment, of that good,
observes in itself some uneasiness or trouble or displeasure till it be done, and this is
what we call desire.

W. von Leyden, John Locke: Essays on the Law of Nature (Oxford: Clarendon Press, 1954), 265-72.

For the importance of this long journal entry, see ibid. 264.

Psychological hedonism as such gives a certain general characterization of the
objects of desire; but it leaves a good deal undetermined. For

end p.120

in itself it says nothing about whether or not an agent always seeks the greatest
possible balance of pleasure over pain, taking future as well as present experiences
into account. This was something about which Locke changed his mind between the
first edition of the Essay, published in 1690, and the second edition of 1694, the latter
being echoed by the third and fourth. Between the first and second editions he
abandoned the doctrine that the will was always determined by 'the greater good in
view'.
To return then to the Enquiry, what is it that determines the Will in regard to our
Actions?

And that upon second thoughts I am apt to imagine is not, as is generally

supposed, the greater good in view: But some (and for the most part the most
pressing) uneasiness a Man is at present under. This is that which successively
determines the Will, and set us upon those Actions, we perform. This Uneasiness we
may call, as it is, Desire: which is an uneasiness of the Mind for want of some absent
good.

Locke, Essay, ii. xxi. 31, pp. 250-1.

And again:
It seems so establish'd and settled a maxim by the general consent of all Mankind,
That good, the greater good, determines the will, that I do not at all wonder, that
when I first publish'd my thoughts on this Subject, I took it for granted; and I imagine,
that by a great many I shall be thought more excusable, for having then done so,
than that now I have ventur'd to recede from so received an Opinion. But yet upon a
stricter enquiry, I am forced to conclude, that good, the greater good, though

apprehended and acknowledged to be so, does not determine the will, until our
desire, raised proportionably to it, makes us uneasy in the want of it.

Ibid. ii. xxi. 35, pp. 252-3.

These two quotations come from the considerable material added by Locke in the
second edition of the Essay and repeated with minor changes in all subsequent
editions. In this emendation, section 28 to 60 of book ii, chapter xxi, replace sections
28-38 of the first edition, and although much of the text of the first edition appears in
the later editions, it appears in a subsidiary role. Philosophically speaking, the
change is significant. For Locke now introduces and discusses at length the subject
of desire, which had been barely mentioned in the first

end p.121

edition.

See Locke, Essay, Nidditch's note to p. 248, line 8.

Moreover, he now gives to desire the principal role in the determination of the will.
Not that he goes back on his psychological hedonism: what happens is rather that
desire

is introduced as a piece of intermediary mechanism, operated on by present

pleasures and pains and by the thought of future pleasure and pain, and itself having
the power to produce volition or will.
By this means Locke hopes to solve the problem known since antiquity as the
problem of akrasia or incontinence. In the first edition he had tried to explain the
problem away in terms of mistakes about the good, saying in words reminiscent of
Plato's Protagoras that present pleasures and pains appear larger than those of the
future on account of their proximity.

Plato, Protagoras 356c4-8, 357a5-b3.

But in later editions he admits that men may indeed know the better and choose the
worse:

Locke, Essay, ii. xxi. 35, p. 254.

Let a Man be never so well perswaded of the advantages of virtue . . . yet till he
hungers and thirsts after righteousness

; till he feels an uneasiness in the want of it,

his will will not be determin'd to any action in pursuit of this confessed greater good;
but any other uneasiness he feels in himself, shall take place, and carry his will to
other actions.

Ibid. ii. xxi. 35, p. 253. There is, however, a different view, or at least a different use of the term

'desire', to be found in Some Thoughts concerning Education, a work which Locke put together, from
notes written nearly a decade before, at the time when he was revising the Essay for the second
edition. There Locke opposes desire to reason, apparently identifying desire with immediate inclination
or appetite. See Some Thoughts concerning Education, sects. 33 and 38; James L. Axtell, The
Educational Writings of John Locke

(Cambridge: Cambridge University Press, 1968), 138 and 143.

Desire, Locke tells us, is uneasiness: the idea of desire is a modification of the simple
idea of pain: a modification because the idea of desire implies a sufficient degree of
pain to determine the will to action.
The first thing we must ask about this newly stressed mechanism is whether Locke
thought it the universal determinant of the will. He explicitly says in several places
that without uneasiness there is no volition, as in the passage last quoted. And
although there is a sentence in which he says that 'the will seldom [my italics] orders
any action, nor is

end p.122

there any voluntary action performed, without some desire accompanying it', the
context suggests that he is here contrasting the operation of 'uneasiness' alone with
'uneasiness' forming part of passions such as anger or jealousy.

Ibid. ii. xxi. 39, pp. 256-7.

There is, however, a stronger reason for saying that Locke does not consistently
affirm that volition is always caused by desire. For in book ii, chapter ii, section 30,
where he wants to distinguish desire and will, he actually gives examples in which he
says 'the Will and Desire run counter'.
A Man, whom I cannot deny, may oblige me to use persuasions to another, which at
the same time I am speaking, I may wish may not prevail on him. In this case, 'tis
plain the Will and Desire run counter. I will the Action, that tends one way, whilst my
desire tends another, and that the direct contrary. A Man, who by a violent Fit of the
Gout in his Limbs, finds a doziness in his Head, or a want of appetite in his Stomach
removed, desires to be eased too of the pain of his Feet or Hands (for where-ever
there is pain there is a desire to be rid of it) though yet, whilst he apprehends, that
the removal of the pain may translate the noxious humour to a more vital part, his will
is never determin'd to any one Action, that may serve to remove this pain.

Ibid. ii. xxi. 39, p. 250.

In consistency Locke should not have suggested that desire can tend in one direction
and will in another, but rather that there may be two desires, of which one determines
the will to a movement which the other opposes. What he needs to show, in order to
prove that will and desire are distinct, is that there can be a desire without a
corresponding action, which does not of course imply that there can be actions
without a corresponding desire.
Locke's usual line from the second edition forward is that volition and hence action
are universally determined by uneasiness, that is, by desire. The only important rider
that he adds is that when several desires have arisen, 'the greatest and most
pressing' does not always determine the will. For the mind has the power to suspend
the execution and satisfaction of any of its desires, giving us 'opportunity to examine,
view, and judge, of the good or evil [i.e. the effect on our happiness] of what we are
going to do'.

Ibid. ii. xxi. 47, p. 263.

end p.123

To support his account of the antecedents of action Locke has to try to show that
whenever we act, or want to act, the volition is triggered by 'uneasiness'. In book ii of
the Essay, chapter xxi, from section 31 onward, this is just what he sets out to do. I
think we can disentangle his arguments more or less as follows.
Locke wants to show that the object of desire is always pleasure, either positive
enjoyment or else the negative pleasure of 'indolency' or relief from pain. He also
wants to show, however, that it is only when we are uneasy at the absence of
pleasure that we have a desire and are thereby moved to action. So he has to
explain how uneasiness comes into the picture in every instance in which we try to
gain pleasure or avoid pain. He must consider the cases in which we wish to
continue present pleasure or get rid of present pain, and also the determination of the
will by the ideas of future pleasures and pains.
Where present pain is concerned he seems to see no difficulty, clearly believing that
this pain, being itself an uneasiness, itself represents desire. And he also insists that
the removal of a present pain is necessarily a condition of our happiness.

John Colman, John Locke's Moral Philosophy (Edinburgh: Edinburgh University Press, 1983), 217

stresses this point.

What is more difficult for him is to say where uneasiness comes in when we want a
present pleasure to continue, and Locke is here driven to the expedient of saying that
we are anxious at the thought that the pleasure will end. 'So that even in joy it self,

that which keeps up the action, whereon the enjoyment depends, is the desire to
continue it, and fear to lose it. . .'.

Locke, Essay, ii. xxi. 39, p. 257.

Future pleasures and pains he deals with in the first place by simply asserting (as in
the passage about thirsting after righteousness which was quoted earlier) that until
we feel uneasy at the prospect of pleasure sacrificed or penalty incurred, we shall not
be turned away from present enjoyments and ease. But he also relies heavily on
examples that seem to bear out his point.
. . . let a Drunkard see, that his health decays, his Estate wastes; Discredit and
Diseases, and the want of all things, even of his beloved Drink, attends him in

end p.124

the course he follows: yet the returns of uneasiness to miss his Companions; the
habitual thirst after his Cups, at the usual time, drives him to the Tavern. . . .'Tis not
for want of viewing the greater good . . . but when the uneasiness to miss his
accustomed delight returns, the greater acknowledged good loses its hold, and the
present uneasiness determines the will to the accustomed action. . .

Ibid. ii. xxi. 35, pp. 253-4.

Such examples cannot really show the nature of desire to be uneasiness, and one
has to admit that Locke's theory is confusing and confused. Nevertheless, Locke
shows his genius in fastening on a connection that is deep, and hard to understand.
For who can deny that there is some very interesting conceptual connection between
pain

, uneasiness, and desire? What Locke says cannot be right; but one may doubt

whether anyone else has yet been able to give the true account of this fascinating
subject.
The important question from our present point of view is how Locke's theory of action
is related to his moral theory, and it is to this that I now turn. As a moral philosopher
he has had very little lasting influence; and this is perhaps not surprising. His moral
theory is a very curious blend of hedonism and rationalism, held together only by an
explicit and often repeated appeal to the existence of God. As we shall see, his
account of the determination of action seemed to force him towards something of this
kind.

Von Leyden wrote that 'Locke's hedonism and certain other views held by him in his later years

made it difficult for him to adhere wholeheartedly to his doctrine of natural law' (Essays on the Law of
Nature

, 14). John Colman, however, sees no inconsistency here, and in my opinion he is right. There

is no reason why one should not be a rationalist about moral knowledge and a hedonist about the
motive to moral action, so long as one is ready to rely on God to annex pleasure to the right actions
either in this world or the next. See Colman. John Locke's Moral Philosophy, 235.

In the Essay concerning Human Understanding Locke claims that moral propositions
can be known with certainty. He is emboldened to claim the status of a science for
morality largely by the similar way in which his theory of knowledge treated moral and
mathematical ideas. Unlike the ideas of material substances such as gold, these
ideas are

end p.125

'adequate'. There is, in such cases, no reference to unknowable 'real essences'
independent of the mind, and so we can have demonstrative knowledge in morals as
we can in mathematics. In the fourth book of the Essay Locke gives several
examples of moral propositions which can be known with certainty (by intuition and
demonstration), as for instance that 'Where there is no Property, there is no
Injustice

Locke, Essay, iv. iii. 18, p. 549.

or 'Murther deserves Death.'

Ibid. iv. iv. 8, p. 566.

Complex ideas such as that of property or murder could be analysed like those of
mathematics. Since such ideas were 'their own archetypes', we could know the real
and nominal essences of property or murder at the same time.

See e.g. ibid. iv. xii. 7 and 8, pp. 643-4.

Locke denies, moreover, that these moral propositions are merely verbal, since the
same ideas with a different name annexed to them would still carry the same
implications.

Ibid. iv. v. 9, pp. 566-7.

Nevertheless, he believed that the examples which he gave of demonstrable moral
propositions were of very little use without a further demonstration which would show
why anyone had an obligation to act in one way rather than another; and this, he
thought, required a knowledge of the laws which God had given to men. In the short
piece called 'Of Ethics in General' published by Lord King in his Life and Letters of
John Locke

, Locke had said that all the knowledge of virtue and vice which man had

attained by the analysis of the complex ideas of morality found in different societies
would amount to little unless we could 'show the inferments that may draw us to
virtue and deter us from vice'.

'Of Ethics in General', sect. 6, in King, Life and Letters, 2nd edn., ii. 127; 3rd edn., 310. Von Leyden

dates this paper to the late 1680s, when Locke was organizing chapters for the first edition of the
Essay

Those who only give definitions
whilst they discourse ever so acutely of temperance or justice, but show no law of a
superior that prescribes temperance, to the observation or breach of which law there
are rewards and punishments annexed, the force of morality is lost, and evaporates
only into words, disputes, and niceties. . . . Without showing a law that commands or
forbids them, moral goodness will be but an empty sound, and those actions which
the schools here call virtue or vice, may by the same authority be called contrary
names in another country; and if there be nothing more than their decisions and
determinations in the case,

end p.126

they will be still nevertheless indifferent as to any man's practice, which will by such
kind of determinations be under no obligation to observe them.

'Of Ethics in General', sect. 9, in King, Life and Letters, 2nd edn., ii. 129-30; 3rd edn., 311-12.

Looking for the source of obligation Locke finds it in laws of various kinds, which he
classifies under the headings of divine law, the civil law, and 'the Law of Opinion or
Reputation

Locke, Essay, ii. xxviii. 7, p. 352.

The sanctions of each kind of law can provide motives to good behaviour, and in the
Essay

he calls them all 'Moral Rules, or Laws'.

Ibid. ii. xxviii. 6, p. 351.

Good and Evil, as hath been shewn . . . are nothing but Pleasure or Pain, or that
which occasions, or procures Pleasure or Pain to us. Morally Good and Evil then, is
only the Conformity or Disagreement of our voluntary Actions to some Law, whereby
Good or Evil is drawn on us, from the Will and Power of the Law-maker; which Good
and Evil, Pleasure or Pain, attending our observance, or breach of the Law, by the
Decree of the Law-maker, is that we call Reward and Punishment.

Ibid. ii. xxviii. 5, p. 351.

The terms 'virtue' and 'vice' are, he observes, applied to actions merely to mark their
conformity with a particular set of mores. Nevertheless, the law which God has given

to man 'whether promulgated to them by the light of Nature, or the Voice of
Revelation'is
the only true touchstone of moral Rectitude; and by comparing them to this Law, it is,
that Men judge of the most considerable Moral Good or Evil of their Actions; that is,
whether, as Duties, or Sins, they are like to procure them happiness, or misery, from
the hands of the Almighty.

Ibid. ii. xxviii. 8, p. 352.

And earlier he had distinguished natural and moral good, saying that 'moral good . . .
is that which produces pleasure of a particular kind, namely the pleasure with which
God rewards certain acts which he considers desirable'.

'Of Ethics in General', sect. 8, in King, Life and Letters, 2nd edn., ii. 128; 3rd edn., 311.

To discover our moral obligations we must, therefore, discover the will of God, and as
the words quoted above suggest, Locke is uncertain in his

end p.127

mind as to whether we can know our obligations by the light of reason or not.

For discussion of this question, see von Leyden, Essays on the Law of Nature, 51-8, and P. J.

Abrams, John Locke: Two Tracts on Government (Cambridge: Cambridge University Press, 1967),
85-6.

Yet in the Essay, book iv, he insists on placing morality
amongst the Sciences capable of Demonstration

: wherein I doubt not, but from self-

evident Propositions, by necessary Consequences, as incontestable as those in
Mathematicks, the measures of right and wrong might be made out, to any one that
will apply himself with the same Indifferency and Attention to the one, as he does to
the other of these Sciences.

Locke, Essay, iv. iii. 18, p. 549.

Immediately before this passage Locke had written of 'the Idea of a supreme Being,
infinite in Power, Goodness, and Wisdom, whose Workmanship we are, and on
whom we depend; and the Idea of our selves, as understanding, rational Beings'

Ibid.

as the elements from which this demonstration might proceed. And the same kind of
suggestion is made, in an amplified form, in a passage from the Essays on the Law
of Nature

, written some thirty years before the Essay concerning Human

Understanding

, and first printed in W. von Leyden's translation from the Latin, in

1954.

'But what it is that is to be done by us can be partly gathered from the end in view for all things. For

since these derive their origin from a gracious divine purpose and are the work of a most perfect and
wise maker, they appear to be intended by Him for no other end than His own glory, and to this all
things must be related. Partly also we can infer the principle and a definite rule of our duty from man's
own constitution and the faculties with which he is equipped. For since man is neither made without
design nor endowed to no purpose with these faculties which both can and must be, employed, his
function appears to be that which nature has prepared him to perform'(von Leyden, Essays on the Law
of Nature

, 157).

It seems that Locke continued to believe that it would be possible to produce a
demonstrative science of morality, and that he even hoped to derive moral laws such
as 'Love thy neighbour as thyself' from the existence of God and from human nature.
But although he was asked to do so by Molyneux and others, he never did produce
anything of the kind.
Molyneux wrote:
One thing I must needs insist on to you, which is, that you would think of Obleidging
the World with a Treatise of Morals, drawn up according to the

end p.128

Hints you frequently give in Your Essay, Of their Being Demonstrable accord ing to
the Mathematical Method.

Letter of 27 Aug. 1692, The Correspondence of John Locke, ed. E. S. de Beer (Oxford: Clarendon

Press, 1976), vol. iv, letter 1530.

And Locke replied:
Though by the view I had of moral ideas, whilst I was considering that subject, I
thought I saw that morality might be demonstratively made out, yet whether I am able
so to make it out, is another question. Everyone could not have demonstrated what
Mr. Newton's book hath shewn to be demonstrable:but to shew my readiness to obey
your commands, I shall not decline the first leisure I can get to employ some thoughts
that way unless I find what I have said in my essay shall have stir'd up some abler
man to prevent me, and effectually do that service to the world.

Letter of 26 Sept. 1692, ibid., vol. iv, letter 1538.

Locke's moral theory was flawed by this failure to do what he claimed could be done,
and even more by his need to appeal to theology to explain moral obligation and the
motivation to act morally. He himself said:
That Men should keep their Compacts, is certainly a great and undeniable Rule in
Morality: But yet, if a Christian, who has the view of Happiness and Misery in another
Life, be asked why a Man must keep his Word, he will give this as a Reason:
Because God, who has the Power of eternal Life and Death, requires it of us. But if
an Hobbist be asked why; he will answer: Because the Publick requires it, and the
Leviathan

will punish you, if you do not. And if one of the old Heathen Philosophers

had been asked, he would have answer'd: Because it was dishonest, below the
Dignity of a Man, and oposite to Vertue, the highest Perfection of humane Nature, to
do otherwise.

Locke, Essay, i. iii. 5, p. 68.

Locke is, therefore, vulnerable to the criticism brought against him, or at least against
doctrines such as his, by Richard Price in A Review of the Principal Questions in
Morals

first published in 1758.

Price there wrote:
Those who say, nothing can oblige but the will of God, generally resolve the power of
this to oblige to the annexed rewards and punishments. And thus, in reality, they
subvert entirely the independent natures of moral good and evil;

end p.129

and are forced to maintain, that nothing can oblige, but the prospect of plea sure to
be obtained, or pain to be avoided. If this be true, it follows that vice is, properly, no
more than imprudence; that nothing is right or wrong, just or unjust, any farther than it
affects self-interest; and that a being, independently and completely happy, cannot
have any moral perceptions. . . .
But to pursue this point farther; let me ask, would a person who either believes there
is no God, or that he does not concern himself with human affairs, feel no moral
obligation

, and therefore not be at all accountable? Would one, who should happen

not to be convinced, that virtue tends to his happiness here or hereafter, be released
from every bond of duty and morality? Or would he, if he believed no future state,
and that, in any instance, virtue was against his present interest, be truly obliged in
these instances, to be wicked?—These consequences must follow, if obligation
depends entirely on the knowledge of the will of a superior, or in the connection
between actions and private interest.

Richard Price, A Review of the Principal Questions in Morals, ed. D. D. Raphael (Oxford: Clarendon

Press, 1974); also in D. D. Raphael, British Moralists (Oxford: Clarendon Press, 1969), ii. 162-3.

In spite of criticisms such as this, Locke's idea of a demonstration of morality
received some favourable notice in the eighteenth century.

See Kenneth MacLean, John Locke and English Literature of the Eighteenth Century (New

Haven:Yale University Press, 1936), 162 ff.

Nowadays, however, it seems to be generally agreed that Locke's moral theory is not
of the same great interest as his political theory or his views on such topics as
substance, personal identity, and the perception of the external world. Hence, no
doubt, the fact that in spite of the interest and vast influence of his political
philosophy, Locke's moral philosophy is now rather little read.
By contrast, modern readers are as much concerned with Hume's moral writings as
with his epistemology. His moral theory is extensively and intensively studied, and if
one had to cite the foremost influence on contemporary ethics one might reasonably
name David Hume. Hume's position is therefore quite different from Locke's so far as
influence is concerned. Moreover, they have sharply contrasting moral theories. In
ethics Locke was a rationalist, claiming as we have seen that morality could be a
demonstrative science on a par with mathematics. Hume stood on the other side of
the great divide that came to separate moral

end p.130

philosophers in the eighteenth century into two schools: the adherents of reason in
ethics and of moral sense.

Bishop Butler's attempt to compromise is expressed in the sentence from his Dissertation of the

Nature of Virtue

(1726) in which he refers to 'a moral faculty: whether called conscience, moral reason,

moral sense, or divine reason; whether considered as a sentiment of the understanding, or a
perception of the heart, or, which seems the truth, as including both'. The Works of Joseph Butler, ed.
W. E. Gladstone (Oxford: Clarendon Press, 1896), i. 399; Butler's Fifteen Sermons, and, A
Dissertation

, ed. T. A. Roberts (London: SPCK, 1970).

He derided Wollaston's attempts to show immorality as a kind of falsehood, and
argued against those who like Samuel Clarke saw moral judgement as the
perception of eternal and immutable relations of fitness which the intellect could
grasp.

William Wollaston, 1660-1724; Samuel Clarke, 1675-1729. For criticism of Wollaston, see Hume,

Treatise

, iii. i. 1, p. 461. For his opposition to theories such as Clarke's, see esp. iii. i. 1, pp. 463-70.

In his own moral epistemology Hume insisted that morality was 'more properly felt
than judg'd of',

Ibid. iii. i. 2, p. 470.

and said that we could never understand the practical nature of morality until we saw
that virtue aroused a feeling of pleasure in our minds. In his theory of the 'artificial
virtues'Hume developed a far more sophisticated system than did eighteenth-century
'sentimentalists'such as Shaftesbury and Hutcheson, who had been content to give
the sentiment of benevolence as the mainspring of moral action; and he insisted that
the peculiarly moral sentiments arose only on an objective and disinterested view of
'qualities useful or agreeable to ourselves or others'.

Anthony Ashley Cooper, third Earl of Shaftesbury, 1671-1713; Francis Hutcheson, 1694-1746.

Nevertheless, in the end, it was by feeling not thinking that we distinguished virtue
and vice.

See Hume, Treatise, iii. i. 2, passim, and An Enquiry concerning the Principles of Morals (1751),

app. i, 'Concerning Moral Sentiment'.

Given this sharp contrast between the moral philosophies of Locke and Hume, it may
seem strange to link their names as legators to modern theories of ethics.
Nevertheless, this is just what I meant to do.

Like his predecessor, Hume had a philosophical theory of the antecedents of action.
Like Locke, he put out a general deterministic thesis, saying that men's actions were
invariably caused by 'their motives, temper and situation'.

Hume, Treatise, ii. iii. 1, p. 404.

But he too went beyond such indefinite statements of determinism, which left the
causes of action more or less open,

end p.131

and told a tale about the specific and universal antecedents of voluntary action. For
Locke this invariable antecedent had been 'an uneasiness'. Hume rejected this idea
and put forward a suggestion that gave him more room to manoeuvre; he said that
the active element in our psychological make-up, ultimately responsible for all our
intentional actions, was 'passion', and this was a more compendious term than
Locke's 'desire'.
Hume's theory of reason and the passions, with reason 'perfectly inert' and the
passions active, with reason the slave and passion the master, is too well known to
need description here.

See esp. Hume, Treatise, ii. iii. 3, pp. 413-18; and iii. i. 1, pp. 457-9.

It was of course a pillar of his anti-rationalist, sentimentalist morality; for how, he
asked, could anything as cool and detached as reason have the active tendency that
so clearly belonged to morality? How could reasoning, whether about ideas or about
matters of fact, give that impetus towards one end rather than another on which all
human action must be based? The import of this theory and the connection between
Hume's psychology and his ethics is easy to see, because Hume put the link
between moral judgement and action at the centre of his moral philosophy and made
it one of his grounds for the rejection of rationalism in ethics.
According to the principles of those who maintain an abstract rational difference
betwixt moral good and evil, and a natural fitness and unfitness of things, 'tis not only
suppos'd, that these relations, being eternal and immutable, are the same, when
consider'd by every rational creature, but their effects are also suppos'd to be
necessarily the same; and 'tis concluded they have no less, or rather a greater,
influence in directing the will of the deity, than in governing the rational and virtuous
of our own species. These two particulars are evidently distinct. 'Tis one thing to
know virtue, and another to conform the will to it. In order, therefore, to prove, that
the measures of right and wrong are eternal laws, obligatory on every rational mind,
'tis not sufficient to shew the relations upon which they are founded:We must also
point out the connexion betwixt the relation and the will; and must prove that this
connexion is so necessary, that in every well-disposed mind, it must take place and
have its influence; tho'the difference betwixt these minds be in other respects
immense and infinite.

Ibid. iii. i. 1, p. 465.

end p.132

Hume's theory of action itself is, however, complex and in places even its outline is
blurred. What did he mean by a 'passion'? What was the relation between passion
and desire? What part, if any, did pleasure and pain play in the determination of
volition? Was Hume, like Locke, a psychological hedonist? How did he think that men
were motivated when they acted morally?
Some of these questions can be answered with confidence. For Hume a passion
may either be an 'aversion or propensity', in which case it is itself a desire, and a
motive to action, or it may be some other 'sentiment' or 'emotion' or 'feeling' such as
love or hatred or pride. The second kind of passion can cause action, by a process

which Hume describes at great length; but as critics have noted Hume does not think
that particular desires are involved in the idea of, for instance, love and hate.

See Patrick Gardiner, 'Hume's Theory of the Passions', in David Hume, ed. David Pears (London:

Macmillan,

1963

), and Páll S. Árdal, Passion and Value in Hume's Treatise (Edinburgh: Edinburgh

University Press, 1966), ch. 3.

Moreover, there are some passions, such as hope, that have no special connection
with action. It is the two passions of aversion and propensity that are the universal
determinants of action.
Beyond characterizing desire in terms of propensity and aversion, Hume does not tell
us what kind of experience desiring is, and it would of course have been against his
principles to attempt to convey by description the content of the simple impression
we receive by introspection and can know only through this experience. But it is clear
that all connotation of tempestuousness has gone from the word 'passion' as Hume
uses it, particularly when he is thinking about desire. As is well known, he even goes
so far as to speak of 'certain calm desires and tendencies, which, tho' they be real
passions, produce little emotion in the mind', and are 'more known by their effects
than by the immediate feeling or sensation', giving as examples 'benevolence and
resentment, the love of life, and kindness to children'.

Hume, Treatise, ii. iii. 3, p. 417. As Árdal points out (Passion and Value, 94), a calm passion is one

so experienced on most, not necessarily all, occasions.

The idea that some desires may be known by their effects rather than by a state of
mind revealed to introspection is, clearly, a point on which Hume differs from Locke,
and I shall have occasion to say more about this later on.

end p.133

So far Hume's doctrine is fairly clear; but interpretation is more difficult when we try to
say how he conceived the connection between propensity and aversion on the one
hand and pleasure and pain on the other.

Hume talks more about pleasure, whereas Locke is driven by his theory of the nature of desire to

place the emphasis on pain.

It is a much-debated question whether Hume was some kind of psychological
hedonist or not.

See Árdal, Passion and Value, 69-79.

What is certain is that he did not think that all action was self-interested. He
disapproved of Hobbes and Locke for their 'selfish system of morals',

Hume, Enquiry concerning the Principles of Morals, ed. L. A. Selby-Bigge, 2nd edn. (Oxford:

Clarendon Press, 1902), app. ii, 296.

and he asks in the final paragraph of the essay on self-love which forms Appendix II
to the Enquiry concerning the Principles of Morals why there should be difficulty in
conceiving that 'from the original frame of our temper, we may feel a desire of
another's happiness or good, which, by means of that affection, becomes our own
good, and is afterwards pursued, from the combined motives of benevolence and
self-enjoyment'.

Ibid. 302.

Thus Hume denies that everything we do is actuated by self-interest. But nor did
Locke think that—if self-interest is to mean the pursuit of perceived maximum
advantage to oneself. We earlier described Locke as a psychological hedonist
because he thought that every agent sought some pleasure for himself, or to avoid
some pain, in whatever he did. And Hume certainly speaks at times as if he believed
this too. Nevertheless, there is a passage in the Treatise in which Hume explicitly
disavows any form of psychological hedonism, saying that 'Beside good and evil, or

in other words, pain and pleasure, the direct passions frequently arise from a natural
impulse or instinct, which is perfectly unaccountable.

Hume, Treatise, ii. iii. 9, p. 439.

Moreover, as Páll Árdal has stressed, even when pleasure or pain is thought by
Hume to be involved in the causal explanation of a desire, pleasure or pain need not
be the end the agent wants to attain.

Árdal, Passion and Value, ch. 3, and esp. p. 73.

Hume was not, we should conclude, a psychological hedonist. Nevertheless,
pleasure and pain come in at a critical point in his moral philosophy. For it is clear
that he believed himself to have solved the problem of

end p.134

linking moral judgement to the will, by insisting that the perception of virtue consisted
of a pleasant sentiment or feeling.'Nothing can be more real, or concern us more,
than our own sentiments of pleasure and uneasiness; and if these be favourable to
virtue, and unfavourable to vice, no more can be requisite to the regulation of our
conduct and behaviour'.

Hume, Treatise, ii. i. 1, p. 469.

Even here, however, it seems most unlikely that what Hume meant was that we are
motivated to act virtuously by the wish to obtain pleasure from the contemplation of
our own virtuous acts. Hutcheson, who was one of the chief influences on Hume's
moral thinking, had argued against philosophers who tried to reduce morality to self-
interest by supposing that we have 'a secret sense of pleasure arising from reflection
upon such of our own actions as we call virtuous, even when we expect no other
advantage from them' and that we are motivated to act virtuously 'to obtain this
pleasure which arises from reflection upon the action'.

Francis Hutcheson, An Inquiry into the Original of our Ideas of Beauty and Virtue (1725), 4th edn.

(Glasgow, 1738; repr. 1772), 96; also in Raphael, British Moralists, vol. ii, sect. 305.

The question of how Hume did understand the role of pleasure and pain in moral
motivation has not so far as I know been satisfactorily answered, and commentators
seem oddly incurious about the matter. Hume seemed to assume that what was
thought of with pleasure tended to be pursued, and perhaps this is a correct idea. It
has appeared, though in a curious form, in our own century in the philosophy of G. E.
Moore, but has not to my knowledge lately received the attention it deserves.

G. E. Moore, Principia Ethica (Cambridge: Cambridge University Press, 1903). ch. 2, sect. 27, p. 42.

We should, perhaps, think again, in the light of Wittgenstein's later philosophy, about expressions such
as 'How good it would be to do such and such'.

I hope that I have by now said enough to show in outline what Locke believed, and
what Hume believed, about the antecedents of voluntary action. I have tried to show
how they differed both in these theories of action and in their moral philosophy; and it
is now time to change to the other tack. For the thesis of this paper is that there is a
common mistake in their philosophies of action. I want to show that neither of them
could give a proper account, within their own theories, of the part that reason plays in
determining human choice.

end p.135

The subject of reason and action, which I have just introduced into this discussion, is
nothing grand or recondite. I am simply referring to the familiar fact that human
beings often do do what they do because it seems the reasonable thing to do.
Sometimes a process of reasoning is involved, as when we decide 'I should do such
and such', or say 'So I'll do such and such' as the conclusion of one of these bits of
practical reasoning. Like speculative reasoning (ordinary non-practical reasoning by
deduction or induction) this kind of practical reasoning has premises and

conclusions. In other cases, however, we act on a reason or for a reason, without
having been through any reasoning. Practical reasoning is then in the background, in
that it may be appealed to if an argument is required.
My charge against Locke, and against Hume, is that the type of explanation of action
which is their stock-in-trade cannot accommodate actions done for a reason. I want
to say that they cannot give a true account of prudential action, when we act in a
certain way for reasons connected with our own future happiness. Nor can they
correctly describe even the most common case of doing one thing because it is
thought that that will be the cause, or enabling condition, of our getting something
else that we want.
Let us first consider prudential action, and think again about Locke's account of the
case where someone renounces present pleasure to avoid later distress. What Locke
says, and must in consistency say, is that we will act prudently only if an 'uneasiness'
arises at the thought of the future evil or loss which present indulgence will bring.
Thus we act as reason bids us only if we find ourselves with the relevant desire. We
have as it were to wait and see whether the thought of the future does or does not
have the effect of arousing the uneasiness in us. And even though Locke thinks that
we have the power to suspend action and again contemplate the future, this is only to
say that we can if we like give the desire more time to appear.
To see how little this kind of theory can accommodate reason-based action, and how
strange it really is, we have only to observe that it must allow as a possibility that
someone who is told that he should do something, like giving up drinking, on
prudential grounds, might reply that

end p.136

he felt no uneasiness at the thought of his future suffering and therefore could not be
expected to do what we advised; as if the necessary mental mechanism was not in
place. And if this does not seem strange enough, it should be remarked that the
same response would be possible in the other case originally mentioned as one in
which reason determines action. For why should it not be given where the reason for
action is of the means-end or instrumental kind, related to a present felt desire? For
example, suppose that someone had a headache (and so far as that goes, an
uneasiness and a Lockean desire to get rid of it) but when told that an aspirin would
relieve his condition he waited to see whether he wanted an aspirin. 'You should take
an aspirin', we say to him, but although he does believe that that would cure the
headache he says that unfortunately he finds in himself no desire to do the thing.

It would, of course, be quite wrong to say that if the desire did not materialize, the agent would

conclude that he did not want to get rid of the headache after all. It is not a question of desires
materializing, but of acting on reasons.

What really happens is just that he says to himself something like 'My headache is
awful. An aspirin would do the trick. So I need to go upstairs to get one', and goes. In
other words, a piece of practical reasoning leads to action. Or perhaps he just goes
upstairs without reasoning, but could give the desire to get rid of the headache as the
reason why

he went. In neither case do we think of inventing a special desire

directed towards going upstairs as an antecedent of the journey. It is enough that
reason tells the agent to go. And the question is why in the other type of case we are
inclined, with Locke, to invoke a desire for our future good when we obey the dictate
of reason which tells us not to drink.
What is at issue here is the supposed requirement for a certain kind of story about
how all action comes about. And it is easy to see that on this point Hume and Locke
are in agreement, in spite of the differences in the exact story that they tell. Where
Locke puts in uneasiness, Hume puts in 'passion' as the necessary antecedent of

action, so that it is in terms of passions that he will have to explain how men act (as
we say) for a reason in curbing their immediate impulses on prudential grounds.
Notoriously, Hume himself even denies that prudential action has any special
connection with rationality.''Tis not contrary to reason', he says,

end p.137

'to prefer even my own acknowledg'd lesser good to my greater.'

Hume, Treatise, ii. iii. 3, p. 416.

The desire for our own good is just one 'passion'among others, and its operation is
mistaken for the working of reason only because, like reason, it is 'calm' and makes
very little disturbance in the mind. So much for Hume's account of the first 'prudential'
example, described above (and rightly described) as one case of the determination of
action by reason.
Turning now to the other example, where the practical reasoning or the reason for
acting was of the instrumental or means-end kind, one finds Hume much more aware
of the problem of fitting this into the general theory than Locke had been.
'Tis obvious, that when we have the prospect of pain or pleasure from any object, we
feel a consequent emotion of aversion or propensity, and are carry'd to avoid or
embrace what will give us this uneasiness or satisfaction. 'Tis also obvious, that this
emotion rests not here, but making us cast our view on every side, comprehends
whatever objects are connected with its original one by the relation of cause and
effect. Here then reasoning takes place to discover this relation; and according as
our reasoning varies, our actions receive a subsequent variation. But 'tis evident in
this case, that the impulse arises not from reason, but is only directed by it.

Ibid. ii. iii. 3, p. 414.

Hume sees that he needs something to be the dynamic antecedent of the action (in
our example, the action of going upstairs) and says that it is the original passion
which is at work, having spread itself, by an association of ideas travelling along the
relation of cause and effect, to the thing that is to be done. And his use of the word
'directed' suggests the image of a force, like a spring of water, or the movement of a
body, which is guided in a particular direction.
Now there is something right in this case about his insistence on the relevance of the
'original emotion', to the action taken, because in this kind of case what we are
talking about is the rationality of taking means to something desired. But I suggest
that he is wrong in thinking that a story was needed about how the desire for the end
energizes the agent to adopt the means. Instrumental reasoning to the fulfilment of a
desire is

end p.138

one of the forms of practical reasoning; and that is all that needs to be said about
how the action comes about. If physiologists or psychologists have some story to tell
about causality on experimental grounds that is, of course, a different matter. But it is
not because we know anything of this kind that we are tempted by the kind of theory
of action given by Locke and Hume.
That we are tempted by this kind of theory—that modern philosophers often make
the same mistake that I have tried to identify in Locke and Hume—is a main thesis of
this paper. But at this point I can imagine that protests will go up. Surely, it will be
said, we are not nowadays inclined to accept that part of the philosophical
psychology of the late seventeenth and early eighteenth centuries which has to do
with the determination of action by desire. Indeed, it may be thought that we accept
very little of it, particularly now that Wittgenstein has taught us to see as problematic

the idea of contents of the mind known to each person only by his own
introspection.

See n. 2 above. As far as I know, Wittgenstein never refers to Locke, but in his later philosophy he

often attacks a theory of language, as well as certain propositions in the theory of mind, which are
almost exactly those which Locke, in particular, put forward.

It is true that contemporary philosophy of mind is radically different from that of Locke
and Hume. Nevertheless, with a few notable exceptions, contemporary philosophers
share the assumption which I have pointed out as being common to Locke and
Hume. To be sure, it is usual now to think of desires as dispositions rather than as
introspectible contents in the mind, and they are therefore treated as underlying
conditions rather than things that occur in the mind immediately before volition. But
as we noticed earlier, Hume thought that some of his calm passions might be 'more
known by their effects than by the immediate feeling or sensation'. The point to be
stressed is that even if Hume had thought of desires as underlying dispositions, this
would not have saved him from the mistake which I am claiming to find in Locke,
Hume, and most present-day philosophers. For an account of how reason affects
action which is in terms of underlying dispositions is still a causal account. Given a
dispositional theory of desire, someone who was told to stop drinking for the sake of
his health

end p.139

could still say that he did not find in himself the desire for health, now conceived as
an underlying condition, a disposition to seek his own good.

Barry Stroud and others seem to me to be mistaken in thinking that Hume's account of desire can

be radically improved by replacing it with a dispositional theory. See Barry Stroud, Hume (London:
Routledge & Kegan Paul, 1977), 167-9.

My point is that where practical reason is concerned no such causal story is needed
to show how action comes about. Practical reasoning, with its counterpart in the
giving of reasons for choosing one course of action rather than another, is a kind of
operation on a par with speculative reasoning. And it is as odd to ask for a causal
account of one as of the other. When someone says that Socrates is mortal on the
ground that all men are mortal and Socrates a man, philosophers do not invent a
causal story about how his belief comes about. Few would suppose, in the case of
speculative reasoning (whether deductive or inductive), that after asserting the
premises the reasoner finds himself with a belief in the conclusion, as if by a causal
mechanism which he might always discover not to have worked. What I am pointing
out is that it would be just as strange, in the case where action is in question, for
someone to say that unfortunately he did not have the desire which would lead him to
act in accordance with reason. One may of course bemoan the fact that what reason
tells us to do (like morning exercises) has so little appeal on its own, and one might
try to increase its appeal by some means or other. But this is not the ordinary way
that practical reasoning works; on the contrary, it is an expedient adopted as a
substitute for practical reason.
What the foregoing pages have shown is that Locke and Hume, however much they
differed in detail, both gave an account of the genesis of voluntary action which is
questionable. For both of them tried to find a type of antecedent which invariably
preceded volition; and the argument of the last few paragraphs has suggested that
this was a mistake that made it impossible for either of them to take proper account
of one most important type of human activity, namely acting on a reason. Looked at
from this point of view the differences between the two philosophers' theories of
action seem slight, while their common presupposition leaps to the eye. Moreover,
the implications of this seventeenth- and

end p.140

eighteenth-century empiricist philosophy are very important indeed. It is no
exaggeration to see here one of the chief influences on the subjectivist, emotivist,
and prescriptivist moral philosophies of our own time, and one of the long-term aims
of the present enquiry is to oppose such theories by striking at their roots.

C. L. Stevenson's Ethics and Language (New Haven: Yale University Press, 1944) is the book that

most explicitly expresses this kind of ethical system; but it is arguable that most moral philosophers in
the analytic school are still deeply influenced by such ideas. G. E. M. Anscombe is a notable
exception, as well as Paul Grice and John McDowell. See G. E. M. Anscombe, Ethics, Religion and
Politics: Collected Philosophical Papers

, vol. iii (Minneapolis: University of Minnesota Press, 1981);

idem, Intention (Oxford: Blackwell, 1957); 2nd edn. (Ithaca, NY: Cornell University Press, 1963); P. H.
Grice, 'Reply to Richards', in R. Grandy and R. Warner (eds.), Philosophical Grounds of Rationality:
Intentions

, Categories, Ends (Oxford: Clarendon Press, 1986); J. McDowell, 'Are Moral Requirements

Hypothetical Imperatives?', Proceedings of the Aristotelian Society, suppl. vol. 52 (1978), 13-30; J.
McDowell, 'Virtue and Reason', Monist, 62 (1979), 331-50. There is also some pioneering work on the
subject of action and desire in Thomas Nagel, The Possibility of Altruism (Oxford: Clarendon Press,
1970), ch. 5.

To try to give some idea of the shift of perspective that would be involved in giving up
the Lockean and Humean point of view, one should remark first of all that what is at
issue is the possibility of a certain kind of explanation of human action. If Locke and
Hume were right, the explanation of every action would involve an antecedent event
or condition such as Locke's uneasiness or Hume's aversion or propensity. If they are
wrong, there is no need to search for any such feature of every situation in which
voluntary action takes place; though of course such things can be taken into account
wherever they really are found. Someone may do something because he is hungry,
because he finds a certain course of action attractive, or because he simply wants to
do that thing. He may, however, do it because he has some reason to do it: it is a
necessary step to something that he now desires; he realizes that he will be in for
trouble later if he does not do it; someone has given him an order; the action falls
under a rule. These are explanations of his action, and in spite of the possibility of
rationalization (in the everyday sense of unconsciously inventing reasons) we should
say that what he gives is normally the correct explanation of why he does what he
does. That is to say that the explanation is in terms of reasons, and any suggestion
that a different account must be given is a philosopher's prejudice. People operate
with

end p.141

practical reason as they operate with speculative reason. It is as wrong in one case
as in the other to invent a parallel process involving psychological causes.
It is easy to see how moral theory is affected by these considerations about the
explanation of action by reasons. For Hume's challenge, which dominated his
account of the perception of virtue and vice, has come also to dominate our own
thoughts about moral judgement. Hume had insisted, as in the famous passage
placed as the motto at the head of this paper, that morality was practical, serving to
produce action, and must be shown as such. Modern theories of moral judgement
tend to take this challenge very seriously, and to meet it by fixing on some
determinant of action, such as feeling or attitude, whose presence is taken to be a
condition of the correct use of moral language; thus introducing a subjective element
into the account. On the principles suggested in the present paper there is, by
contrast, no need to look for any such thing. No such element need come into the
explanation of action as motivated by the thought of reasons. And the way is open for
a simpler account of the necessarily practical character of morals. For undoubtedly
those who have successfully been taught morality see moral considerations as
reasons for action. We do not have to look for something special in the way of 'moral

motivation' to see how it can be that they do things, on many occasions, because
morality so dictates.
Nor is it hard to see on these principles that morality is necessarily practical; which is
to say not that people always act morally (which of course they do not) but rather that
there is a conceptual link between morality and action. To show just how this is true
is a more complex matter, and it is no doubt hard to get it quite right. Perhaps it is
enough here to suggest that a moral system necessarily involves some kind of rules,
and to point out that rule belongs to our armoury of practical linguistic devices. If
someone should insist that it is not necessarily rational to obey any and every rule,
this observation should be welcomed and not taken as any objection to what has
been said. For what I have been trying to do here is not to produce some kind of
justification of morality—of the making and keeping of moral rules—as a rational
human enterprise, in contradistinction to the invention of duelling rules or the more

end p.142

byzantine varieties of etiquette. Nor have I tried to show that someone who acts for a
moral reason acts for a good reason. What has been in question is rather a clear
view of the way in which morality motivates action, and above all a rejection of the
kind of story which, in consistency with their own philosophy of mind, Locke and
Hume had to tell.

end p.143

Nietzsche's

Immoralism

Philippa Foot

In writing about Nietzsche's immoralism I am going to ask a simple question about
him, something that is difficult to do: it is hard to hold onto anything simple in the face
of this determined joker, who loved masks and hidden things, and whose protean
style is sometimes of the most lapidary aphoristic simplicity but often lush and
rhetorical. It has been said that Thus Spake Zarathustra should be read as an opera,
and it is surely a great shame that we never had a rendition by the comedy singer
Anna Russell of those wild journeys between mountain, marketplace, and cave.
Nietzsche thought he could discredit morality; and I want to ask, 'Was he right?' I
think the question should be asked. It is always respectful to ask of a great
philosopher whether what he says is true, and hardly respectful not to ask it. Why do
so many contemporary moral philosophers, particularly of the Anglo-American
analytic school, ignore Nietzsche's attack on morality and just go on as if this
extraordinary event in the history of thought had never occurred? It is true, of course,
that it is hard for those of us who belong to the plain-speaking school of analytic
philosophers to grapple with his work. We are used to ferreting out entailments, and
lines of argument, and building up a theory from individual passages. And I do not
think that one can work on Nietzsche quite like that. The unity of his writings—which
is most remarkable in spite of their amazing richness and many superficial
contradictions—

end p.144

comes from his attitudes, from his daring, his readiness to query everything, and from
his special nose for vanity, for pretence, for timid evasion, and for that drive to
domination which he finally supposed to be the principle of all life.
One must take account of Nietzsche's attitudes; of the contempt he felt for modern
European man, for the 'newspaper-reading' public,

Beyond Good and Evil, 263.

for democracy, for nationalism, for Bismarck and all things German (save for Goethe,
'the exception among Germans'

The Gay Science, 103.

). And account too, of course, of his vituperative attitude to Christianity, which he saw
as the religion of pity and weakness but also, at times, as the beneficially tyrannical
source of spiritualization in man.

Beyond Good and Evil, 188.

One has to remember that Nietzsche was one who wanted to be an affirmer, not a
caviller, who repeatedly praised lightness of spirit, and wrote much about dancing
and laughter. When he put forward his strange theory of the eternal recurrence of all
things—round and round again—this was most significantly a rejection of gloomy
nihilism and a way of saying 'yes' even to his own physically painful, and painfully
lonely, life.
All this, and much more, is needed to interpret Nietzsche. But what, then, can he
have to offer to the descendants of Frege and Russell, of G. E. Moore and
Wittgenstein? What can we ourselves take from the strange Nietzschean symphony
of subjectively interrelated attitudes and beliefs? Even in those matters in which there
is overlap between his interests and ours, can we assume that he is seriously
concerned with the truth? Was it not Nietzsche who saw truth in terms of divergent
'perspectives', and who insisted on questioning the value of truth itself? He said all
this, and meant it. Nevertheless, he saw as a great sign of those things he so much
celebrated, 'strength' and 'life', the ability to face reality as it is. Honesty (Redlichkeit)
was, he wrote, the one virtue that he and other 'free spirits' must take from morality,
that they could not leave behind:
let us work on it with all our malice and love and not weary of 'perfecting' ourselves in
our virtue, the only one left us. . . . And if our honesty should nevertheless grow
weary one day and sigh and stretch its limbs and find us too hard let us dispatch to
her assistance whatever we have in us of devilry.

Ibid. 227.

end p.145

Nietzsche may have thought of even his own views as merely his truths (whatever
exactly that means). But his love of truth was based on one of the strongest things in
him, that is on his contempt for evasive falsification. So in spite of all the discouraging
omens, I want to ask what truth there could be in the doctrine that makes us name
Nietzsche, as he sometimes named himself, 'immoralist'.
Nietzsche's immoralism! A host of problems and many interpretations live together
under this roof. Was he perhaps preaching in favour of a new morality rather than
against morality as such? I think not. Nor was Nietzsche simply a run-of-the-mill
moral relativist. He branded as 'childish' the idea that no morality can be binding
because moral valuations are necessarily different among different nations.

The Gay Science, 345.

So even his arguments for the subjectivity of moral judgement were idiosyncratic. He
saw different moralities as determined by the desires and needs of peoples and
generations: at one time the need to control aggressive individuals when they were
no longer useful in meeting external enemies; in the long reign of Christianity the
desire of the weak and 'misbegotten' to brand themselves as 'good' and those
stronger characters, whom they feared, as 'evil'; in modern Europe the longing of the
mediocre 'to look nobler, more important, more respectable, "divine' ".

Ibid. 352.

Throughout all these changes morality was, Nietzsche insisted, fundamentally a
subterfuge by which the weak—the members of the herd—tried to dress up their
weakness and their fears as 'goodness', a device by which they produced self-doubt
and a bad conscience in those who, as nobles, had once unquestioningly called
themselves good. The 'nobles', the type of the original barbaric Greek and the
Renaissance Man, had called 'inferior' men bad (schlecht) only by contrast to
themselves. The 'inferiors' on the other hand needed to see dangerous men as 'evil'
(böse) so as to see themselves as good.
In suggesting that different moralities were rooted in the different needs, fears, and
desires of different peoples Nietzsche was applying to valuations the
characteristically Nietzschean 'perspectivism': the interpretation by historical
genealogy, and above all by underlying desires, that he applied to all modes of
thought. He applied it particularly to

end p.146

abstract philosophies, which he saw as expressing instincts, needs, and fears rather
than that will-o'-the-wisp, 'pure thought'. Thoughts, he said, 'are the shadows of our
feelings, always darker, emptier, and simpler'.

Ibid. 179.

But there is, of course, something more specific than this in Nietzsche's insistence
that 'there are no moral facts'.

Twilight of the Idols, vii. 1.

This problem of the value of pity and of the morality of pity . . . seems at first sight to
be merely something detached, an isolated question mark; but whoever sticks with it
and learns how to ask questions here will experience what I experienced—a
tremendous new prospect opens up for him, a new possibility comes over him like a
vertigo, fear leaps up, his belief in morality, in all morality, falters—finally a new
demand becomes audible . . . we need a critique of moral values, the value of these
values themselves must

. . . be called in question.

Genealogy of Morals, preface, 6.

Nietzsche says that he is going to query the value of moral values, which suggests
that he has some other value in play. And there is, indeed, a positive side to
Nietzsche's ideology. He is affirming a special kind of aestheticism, and attacking
morality partly on its own ground but partly in the interest of what he calls the
'ascending' type of man. What was to be seen as 'good' was the 'strong', 'fine',
'noble', 'subtle' type of human being. This free and joyous spirit, subjecting himself to
the sternest discipline but accepting no rule from others, was sometimes seen by
Nietzsche as the 'overman', the superman of Nietzschean popular legend: that is as
one who belonged to the future. But actual human beings might be seen as stepping
stones or bridges on the way to this future. The important question to ask about any
man was whether he represented an ascending or descending type. This was the
profound classification, and determined the worth for the particular instance of those
elements of character and action that moralists wrongly thought significant in
themselves. So egoism, for instance, should not be thought of as either bad or good
in all individuals.
The value of egoism depends on the physiological value of him who possesses it: it
can be very valuable, it can be worthless and contemptible. Every individual may be
regarded as representing the ascending or descending line of life.

end p.147

When one has decided which, one has thereby established a canon for the value of
his egoism.

Twilight of the Idols, 33.

Nietzsche thus, very characteristically, saw our common moral classifications as
reflecting reality in a herd-based way that was deleterious to the exceptional man.
What was worst about them, and was common to all morality, was the attempt to
determine the value of any kind of conduct in the case of each and every
person.'Good and evil the same for all, ' he scoffed. There could be no beneficial
rules of conduct.'A virtue has to be our invention, our most personal defence and
necessity: in any other sense it is merely a danger.'

The Antichrist, 11.

And again, " 'Good"is no longer good when one's neighbour mouths it.'

Beyond Good and Evil, 43.

Thus Nietzsche thinks of value as belonging only to a person who has created his
own character in a pattern that cannot be prescribed for others, and it is here that his
shift from a moral to an aesthetic form of evaluation becomes clear. Not surprisingly,
he is writing of what he himself, as a genius of style and image, knew best. Not for
nothing does he say in one place, 'We want to be the poets of our lives.'

The Gay Science, 299.

The discipline that he so much stresses for the creation of a splendid individual
human being is modelled on the discipline of the artist. For an artist, rules would
indeed be beside the point:the goodness of what he or she makes cannot be the
same as the goodness of other artists'work, as if there could be a manual for
producing what is good. This analogy seems to be an essential element in
Nietzsche's aestheticism—in his shift from moral to aesthetic valuation. Theoretically,
it is separate from his perspectivism, since, after all, the absence of rules for artistic
creativity does not entail the subjectivity of aesthetic judgement. But when the
individual himself is both artist and artwork they come together in the fact of his
special 'interpretation' of the world, the interpretation that determines what he sees
as good.
There have been many attempts to see all this as an inspiring call to a kind of joyous
paganism that would leave us with all that is best in morals. Can this be sustained? I
think not, just because of Nietzsche's

end p.148

attack on the universalism in morality. He insists that there are no kinds of actions
that are good or bad in themselves, and this has, it seems, a fatal implication for the
teaching of justice. It is justice—understood as one of the four cardinal virtues and as
having to do with all that one person owes another—that forbids such acts as
murder, torture, and enslavement and brands them as evil, whoever carries them out.
Nietzsche, on the other hand, says that there is nothing good or evil 'the same for all',
and he tells us we must look to see what kind of a person is doing an action before
we can determine its 'value'.
If this implies, as it seems, that not even the most flagrant acts of injustice can be
called evil in themselves, then was Thomas Mann not perhaps right in saying that
Nietzsche had not faced the reality of evil? Mann said in 1947, 'How bound in time,
how theoretical too, how inexperienced does Nietzsche's romanticizing about
wickedness appear . . . today! We have learned to know it in all its miserableness.'

Thomas Mann, Nietzsche's Philosophy in the Light of Contemporary Events (Washington: Library of

Congress, 1947).

Mann was writing, of course, soon after the facts about Belsen and Buchenwald, and
their images, had come to haunt us. So however much the Nazis had had to distort
Nietzsche in order to claim him as one of their prophets, Nazi actions and Nietzsche's

reputation may be linked in the way suggested by Mann; that is, in the way his
treatment of evil has to look to us in the light of what they did.
It may be argued that this is unfair to Nietzsche. It may be pointed out that neither
Hitler nor Stalin were individuals of whom it should be thought for a moment that they
embodied his ideals. J. P. Stern is surely mistaken when he writes, 'No man came
closer to the full realization of self-created "values"than . . . Hitler.'

J. P. Stern, Friedrich Nietzsche (Harmondsworth: Penguin, 1979), 86.

Nietzsche is, after all, vituperative about merely cruel monsters, and while, to be
sure, he praises the (as he says) 'pranksomely' ruthless 'nobles' above the resentful
'herd' Alexander Nehamas seems right to say that they do not need to be seen as his
ideal for all times.
Nietzsche's defenders may, of course, also remind us of what he said about the need
to discipline the passions, which is indeed a central

end p.149

element in his philosophy. For Nietzsche is not at all like Callicles, the immoralist in
Plato's Gorgias, whose ideal is that of the libertine. Nietzsche preaches hardness and
self-mastery. The passions are not to be weakened or extirpated, but used in the
creation (once more one thinks 'it's like the artist's creation') of the self. Moreover he
puts forward a doctrine of the sublimation of the passions (he was one of the first
actually to use the term sublimieren), believing for instance, that the 'drive' of cruelty
could be turned into a desire for truth. It will be said therefore that Nietzsche did not
actually countenance acts of injustice in substituting for morality's canon against such
things as murder and oppression his own prescription of self-creation. Did he
perhaps believe that no one who truly embodied the Nietzschean ideal would ever
find himself in such actions? Might the ideal of self-realization turn out in the end to
be unshocking?
I am sure that something of all this is true, and that one side of Nietzsche would have
welcomed such an accommodation. He speaks of gentleness, in some convincing
passages; and he was himself, I would suppose, for all his insistence on the
beneficial effect of suffering, actually oversensitive to it in others, really experiencing
pity as he notoriously represented it—as 'suffering's contagion'. The character of the
man himself shows too in his heroes and the books he loved. Cesare Borgia was not
a hero of his, in spite of his preference 'even', as he notoriously said, for him over a
mean-spirited member of 'the herd'. True, he admired Napoleon, but said that he was
'half superman half monster'.
Nietzsche's great hero was, it seems, Goethe, whom he praised especially for his
moulding of sensuality and spirit into a harmonious self. And among the literary
works Nietzsche most loved there were not only the novels of Stendhal and
Dostoevsky but also two quiet-mannered books, Eckermann's Conversations with
Goethe

and Emerson's Essays, a book he 'felt at home in' and seems to have kept

by him for much of his life. (One gets interesting light on Nietzsche from both of these
works.)
Nevertheless, there was a side of Nietzsche's deeply pathological psyche that seems
to have gloried in the fact that his immoralism allowed, if done by certain people,
even terrible deeds. Unlike other proponents

end p.150

of self-realization Nietzsche does not say that these acts could never be a sign of
health and of truly 'becoming what one is'. On the contrary he stresses the
fearfulness of his 'revaluation of values'. He insists that he has set out on a journey
over terrifying seas, and, from the time in the early 1880s when he first started to

attack morality, to the end of his working life, one can find passages that stress the
fearfulness of his thought, and seem to license injustice.
In The Gay Science of 1882 he writes,
Hatred, the mischievous delight in the misfortunes of others, the lust to rob and
dominate, and whatever else is called evil belongs to the most amazing economy of
the preservation of the species.

The Gay Science, 1.

And again in the same work,
Some kinds of hatred, jealousy, stubbornness, mistrust, hardness, avarice, and
violence . . . belong among the favourable conditions without which any great growth
even of virtue is scarcely possible. The poison of which weaker natures perish
strengthens the strong nor do they call it poison.

Ibid. 19.

Four years later, in Beyond Good and Evil, he writes that
everything evil, terrible, tyrannical in man, everything in him that is kin to beasts of
prey and serpents, serves the enhancement of the species 'man' as much as its
opposite does.

Beyond Good and Evil, 44.

And in a note from 1887 included in the Nachlass collection The Will to Power:
when one makes men more evil, one makes them better

—. . . one cannot be one

without being the other.—At this point the curtain rises on the dreadful forgery of the
psychology of man hitherto

The Will to Power, 786.

Perhaps these passages are not absolutely decisive. Perhaps Nietzsche is talking
about 'drives' that might be 'enhanced' and 'strengthened' before being sublimated
into harmless actions. But this does not seem at all plausible in the face of his
insistence that his doctrine is a fearful one.
In any case I do not think it should be argued that the virtue of justice

end p.151

can be accommodated within Nietzsche's picture of splendid individuals finding each
his own values and 'his own way'. For there is something in Nietzsche's description of
this 'higher type' of human being that positively tells against it. I mean the way in
which the self-guiding person is described as seeing those whom he counts as
'inferiors'. One simply cannot ignore all that Nietzsche says, approvingly, of the
experience, the feeling, the 'pathos' as he likes to put it, 'of distance', of being not just
apart from, but higher than, those who belong to 'the herd'. Nietzsche says at one
point that contempt is better than hatred, and of course he thinks the idea of equality
utterly despicable.
Now what I wonder is this: whether the practice of justice may not absolutely require
a certain recognition of equality between human beings, not a pretence of equality of
talents but the equality that is spoken of in a passage of Gertrude Stein's when she
says (pretending to be Alice B. Toklas) that she herself had a sense of equality, and
that was why people would help her. 'The important thing . . . is that you must have
deep down as the deepest thing in you a sense of equality.' This is particularly
striking in Gertrude Stein, who was certainly not one to underestimate her own
individuality, talent, or place in literary history. The sense of equality that she is
thinking of must, surely, have to do with thinking that one is always, fundamentally, in
the same boat as everybody else, and therefore that it is quite unsuitable for anyone
to see himself as 'grand'.

Perhaps I am wrong in thinking of this sense of equality as necessary for the practice
of justice. That the two are connected seems, however, to be supported in a certain
passage I once came across in which G. K. Chesterton wrote about Charles Dickens.
Dickens, Chesterton said,
did not dislike this or that argument for oppression:he disliked oppression. He disliked
a certain look on the face of a man when he looks down on another man. And the
look on that face is the only thing in the world that we really have to fight between
here and the fires of hell.

Introduction to the Everyman edition of Oliver Twist (London: J. M. Dent, 1907).

Nietzsche's endless talk about inferiors and superiors, and the way he countenances
some men looking down on others, together with his

end p.152

own readiness to sacrifice—to write off—the 'mediocre', confirms the impression that
justice gets short shrift in his scheme of things: that it is quite wrong to see his
'aesthetic' as taking nothing we think precious from the morality he attacks.
Nietzsche's defenders will rise up, of course, to insist that the 'looking down' that he
speaks of is nothing so crude as that of which G. K. Chesterton speaks. But the
language of contempt is undeniably there. Nietzsche's defenders are like those who
say of Wagner that he is better than he sounds.
To our objections on behalf of justice Nietzsche would, no doubt, reply that what
should be in question is not whether we want to hold on to a moral mode of valuation,
but whether we can do so with honesty. For his contention is that morality is tainted
by certain pious falsehoods that are necessary to it; so that morality, in praising
honesty, sowed the seeds of its own demise. Therefore we do have to ask ourselves
not just what Nietzsche's own system of valuation amounts to but also if morality can
withstand his attack.
What were these falsehoods—the 'errors'—that Nietzsche saw as endemic to
morality?
First there is the belief in free will, which he challenged on the ground that will itself,
as required for either free or unfree will, is non-existent. What we call will is, he said,
in truth nothing but a complex of sensations, as of power and resistance, and it is
pure illusion to think of it as a basis for 'moral responsibility'. Our actions arise not
primarily from conscious motivations but rather from physiological and psychological
factors of which we are unaware.
It follows, Nietzsche thinks, that men are totally innocent, as innocent as anything
else in the world, though this, he says, is something we hate to accept.
Man's complete lack of responsibility for his behaviour and for his nature, is the
bitterest drop which the man of knowledge must swallow if he had been in the habit
of seeing responsibility and duty as humanity's claim to nobility. All his judgements,
distinctions, dislikes have thereby become worthless and wrong: the deepest feeling
he had offered a victim or a hero was misdirected; he may no longer praise, no
longer blame, for it is nonsensical to praise and blame nature and necessity. Just as
he loves a good work of art, but does not

end p.153

praise it, because it can do nothing about itself, just as he regards a plant, so he
must see the actions of men and his own actions.

Human, All Too Human, 107.

The topic of free will and moral responsibility is itself so large that one cannot quickly
assess Nietzsche's idea that there is here an error on which morality is based. But it
may be pointed out that the theory of the will that he attacks would find few defenders
today; and of course few would deny unconscious motivation. Nevertheless moral, as

opposed to aesthetic, evaluation does require some distinction between actions for
which we are responsible and those for which we are not responsible. For moral
evaluation describes a person in terms of virtues such as courage and justice and
charity, and we cannot, of course, ascribe virtues to anyone without knowing first of
all which of the things that he did were intended and which unintentional, and
secondly which of the unintentional actions were due to lack of care, or to ignorance
of that which he could and should have known.
It is not, however, obvious that these distinctions rest on a doctrine of 'moral
responsibility' that Nietzsche is in a position to deny. He is surely wrong in thinking
that we might have to give up thinking in a special way about the goodness of men:
that we should have to relinquish the concept of a virtue as it applies to human
beings and not to plants or to the objects of aesthetic evaluation. The idea of a virtue
might even be the correct starting-point for a solution to the problem of moral
responsibility. For the way in which moral responsibility exists can perhaps be traced
precisely by asking how it enters into the concept of a virtue, as shown by the
irrelevance to virtue of things done accidentally or in (many cases of) ignorance. And
as for unconscious motivation: we might say that this is relevant to moral evaluation
(as when we count a person's deep hidden malice against a claim to the virtue of
charity) without any implication that the subject is 'responsible' for being as he is. So
far from destroying morality, Nietzsche's challenge to the possibility of distinctively
moral evaluation may actually help us to see what it does and does not require.
Second among the 'errors' Nietzsche claims to have found in morality

end p.154

there is the classification of types of actions under the descriptions 'good' and 'bad'.
For Nietzsche's objection to this we must go back once more to his scorn for the
universality in moral judgement, his scorn for its branding of certain kinds of action as
good or bad 'for all'. This was not the commonplace insistence on the relevance of
circumstances to moral good and evil. It was not that objection to absolutism which
Nietzsche had in mind; he meant rather that moral generalization was impossible
because the proper subject of valuation was, instead, a person's individual act. We
were to ask not what is done, but rather by whom it is done. He even said that no two
actions can be the same, meaning, again, that each individual action takes its
character from the character of the one who does it.
His chief defence for this comes, I think, from the sceptical eye that he casts over the
motives of the actions that moralists call good. Thus he points out the vanity that is
behind many acts of 'kindness': the wish to create a good opinion in others by a
kindly deed, so as to be able to buy this good opinion back from them. (As T. S. Eliot
said, 'the endless struggle' to think well of ourselves.) The wish to be a benefactor
was, he said, impertinent in its claims to understanding the one to whom 'good' was
done, and jealous in the desire to possess him. Where moralists find altruism
Nietzsche sees various kinds of egoism, self-mistrust, and fear: above all the desire
to 'live abroad' with others rather than at home with oneself. Under the heading 'The
elevating aspect of our neighbour's misfortune', he says that we gather to bemoan
the ill that has befallen him and 'spend an enjoyable afternoon'. Nietzsche was a
genius at finding hidden motivations, and it is not surprising that Freud found him so
much of a kindred spirit that he deliberately avoided reading Nietzsche until his own
work was well advanced.
It is surprising, however, that Nietzsche thought the discovery of the possibility of
dubious motivation behind, for example, acts of 'kindness' to be a count against the
moral mode of valuation itself. For it is traditional in moral philosophy that actions are

to be judged not only for the type of actions that they are but also as individual acts
done by a particular agent at a particular time. Aquinas, for instance, pointed out that
a concrete act could be spoiled, morally speaking, either by what it was 'in

end p.155

its kind', as for example murder or robbery, or by the motive from which it was done,
using for this latter possibility the example of giving alms 'for the praise of men'. If
Nietzsche extends the range of experience in which the standard of honesty about
motives applies, moralists should not take this amiss.
So far, then, Nietzsche seems to be on strong ground in his psychology, even if
mistaken about the import of his psychological observations. It is not, however,
always so, and the next of the 'errors' he claims to find in morality sees him far out in
a very doubtful field of psychological speculation. For he believed that he could
discern the 'drives' (Triebe) that motivate all human action, and could map their
dependence on one another. He thought he knew, for instance, that 'drives', such as
cruelty, that were branded by moralists as 'evil', were the condition of all 'good'.
Thus, in Beyond Good and Evil he speaks of 'the reciprocal dependence of the
"good" and the "wicked" drives' and the derivation of good impulses from wicked
ones; continuing, in a famous passage, that we 'should regard even the affects of
hatred, envy, covetousness, and the lust to rule as conditions of life, as factors which,
fundamentally and essentially, must be present in the general economy of life (and
must therefore be further enhanced if life is to be further enhanced)'.

Beyond Good and Evil, 23.

This was a favourite thought of Nietzsche's: one that he several times illustrated with
the image of a tree which to flourish had to have its roots in the mud.

The Gay Science, 171.

He saw that his views about 'evil' drives were inimical to morality, because morality
has to set its face against certain desires; and he must surely be right about that. But
whether there is the least warrant for the kind of psychological speculations that
would support this part of Nietzsche's immoralism is quite another matter. In the
theory of 'drives' that finally crystallized into the theory that all 'drives' are contained in
the Will to Power, Nietzsche seems to have fallen into the trap of working a modicum
of psychological observation into an all-embracing theory which threatens to become
cut off from facts that could possibly refute it. Nietzsche saw himself as a wonderful
psychologist, but the truth is that he was partly a wonderful psychologist and

end p.156

partly a mere speculating philosopher far exceeding any plausible basis for his
speculations.
Is no part of Nietzsche's attack on morality, then, convincing? Probably not. It would
be wrong, however, to conclude that we analytic philosophers should leave him
alone. On the contrary, I think that he should shake us up. For his deepest conviction
was that the fact that 'God is dead' (so that nothing is guaranteed to us) could not
leave our faith in morality unchanged.

The Gay Science, 343.

He was particularly scornful of 'philosophers'—he singled out George Eliot—who
were 'fanatics' for morality in spite of their atheism. Nietzsche believed, in effect, that
as the facts of human psychology really were, there could be no such things as
human virtues, dispositions good in any man; and even if he did not prove it, might
he not alert us to the fact that that could be how it is? For if 'God is dead' what
guarantees that there is a human aptitude for the virtue of justice, given that this
requires quite generally that men and women can do certain things—as, for example,

pass up great advantage in refraining from murder or theft and moreover do this in a
certain way

: that is without ulterior motive, false elevation, or bitterness? Wittgenstein

has taught us to see the existence of some things we take for granted as being a
remarkable fact. Should we, perhaps, see the capacity to acquire justice in this light,
as depending on certain general human reactions to teaching, somewhat as it is with
the capacity to learn to talk or to make calculations?
On grounds such as this, one can well believe that analytic philosophers must lose
something if they do not study a philosopher as surpassingly bold and original as
Nietzsche, if only because of his capacity to stretch our philosophical imagination.
And of course if I am right there is also work to be done in criticizing his theories from
the point of view of philosophical argument and truth. This is what I have been just
beginning to do here. In a way it is bound to be a somewhat comical proceeding,
because it has to be carried out at a schematic level that leaves behind all the riches
of Nietzsche's psychological insights and images. So one feels rather like a surveyor
reducing a glorious countryside to contours,

end p.157

or like someone telling the Sirens they are singing out of tune. But that is not to say
that this rather dry philosophical work can be left undone, especially if, as I think,
Nietzschean teaching is inimical to justice. His teaching has been sadly seductive in
the past. Who can promise that it will never be seductive again?

end p.158

Rationality and

Virtue

Philippa Foot

This paper is about the rationality of moral action, and so about a problem that is as
old as Plato but which still haunts moral philosophy today. It is about the rationality of
following morality: of refraining from murder or robbery for instance, and being faithful
in keeping contracts and promises even where this seems to be against our interest
and contrary to what we most desire. The problem of the rationality of morality arises
most obviously over such actions and therefore has to do particularly with the virtue
of justice, because it is here that self-interest and morality often seem to clash. Then
Reason may represent itself as on the side of self-interest and the fulfilment of
present desire; so unless it can be shown that acting justly is a necessary part of
practical rationality, cynics like Thrasymachus will always say that there is no good
reason to pass up an advantage for the sake of acting justly, and plenty of reason not
to pass it up.
A great deal is at stake here. For if Thrasymachus is right the just person will often
act irrationally, and we take it for granted that there is something wrong with doing
that; so that someone who acts irrationally does not act well. And because it is in the
concept of a virtue that

end p.159

it makes its possessor's action good, the status of justice as a virtue is being
questioned by Thrasymachus, as he himself said with a sneer.

Plato, Republic, 348c2-d3.

By what strategy should this problem about the rationality of morality be addressed?
It may be at this point, when strategy is determined, that a crucial error is very often
(even usually) made. It is assumed that our task is to reconcile the requirements of
justice with a preconceived theory of practical rationality; as if we had to show that
despite appearances it always is in our interest to act justly, or that that is the way to
get what we presently want most. Such attempts have, in my opinion, always

foundered on the case of the 'tight corner', where someone might be able to save his
life only by ruthlessly ignoring other people's interests and rights. This kind of
example obviously raises difficulties if practical rationality is taken to be self-
interested action. And other more strictly Humean doctrines are not in any better
case. If, for instance, practical rationality consisted in the pursuit of the maximum
satisfaction of present desires, there would still be, at best, a contingent coincidence
between rational action and morality, and even that would be hard to make out.
It is natural therefore to ask whether the intractability of this problem may not be due
to something in the strategy by which we try to solve it. In pursuit of that thought I
want, in a sense, to turn the problem on its head: to start out not from a theory of
practical rationality but from the idea that justice is a virtue. This may sound like the
most outrageous question-begging, given that the status of justice is exactly what is
in doubt; but there can be nothing wrong with exploring such a hypothesis, and I shall
now do that.
To discover the implications of the idea of justice as a virtue we must ask what a
virtue is. It is said to be an 'excellence' but that does not take us very far when we
want to think not about the virtue of sharpness in a knife, or meticulous carefulness in
a scholar, but about the putative virtues of human beings, and of human beings as
such. These 'excellences', which we often call moral virtues, are supposed to make
their possessors good. But in what respect? The answer to this is somewhat

end p.160

controversial, since some say that virtues 'rectify' passions as well as actions, and
others do not. I shall bypass this controversy by pointing out that it is agreed on all
sides that virtues, whatever else they do, must make their possessors'actions good.
So we shift our attention to the concept of good action. An action is something done
by a human being, which may come to grief for irrelevant reasons like non-culpable
ignorance or lack of physical strength, but can also be judged as bad as an action.
So what exactly does a virtue, in the sense in which justice is supposed to be a virtue
do? What distinguishes a just person from one who is unjust? The fact that he keeps
his contracts? That cannot be right, because circumstances may make it impossible
for him to do so. Nor is it that he saves life rather than killing innocent people, for by
mischance it may be the latter rather than the former that he does.'Of course',
someone will say at this point, 'It is the just person's intention not what he actually
brings about that counts'. But why not say, then, that it is the distinguishing
characteristic of the just that for them certain considerations count as reasons for
action

? (And as reasons of a certain weight.) Will it not be the same with other

virtues, as, e.g., of charity, courage, and temperance? Those who possess these
characteristics possess them by virtue of the fact that they recognize certain
considerations as powerful, and in many circumstances compelling, reasons for
acting, and that their actions reflect their views.
The description 'just' as applied to a man or woman speaks of how it is with them in
respect of the acceptance of a certain group of considerations as reasons for action.
So if justice is a virtue, it is this that justice makes to be good: the just are good in
that they recognize and act on certain reasons. Which is to say—since it can hardly
be one thing to be practically rational in a certain area and another thing to be as one
should be there so far as the recognition and following of reasons is concerned—that
if justice is a virtue this is because it takes care of a certain area of practical
rationality. Possession of a virtue makes a person do well in respect of acting on
reasons, and justice would make him do well where the reasons are about, e.g.,

promises, property rights, or respect for life. This is not a surprising conclusion, given
that it was recognized earlier that the status of justice as a virtue was at stake in the

end p.161

dispute with Thrasymachus. All we have done in the preceding paragraphs is to spell
out the reason why if justice is a virtue just action must be rational action, and so why
if the latter is not the case the former cannot be true.
It seems, therefore, that we have arrived at a point at which it is crucial to decide
whether justice is a virtue or not. Does someone act well—is his action so far forth
good—if he is just? This is a weighty question, to which I suppose there is as yet no
satisfactory answer. But it will be useful here to advert to some considerations
already familiar from the work of various moral philosophers.
In the first place I should like to pick up Peter Geach's use of the idea of 'good' as an
'attributive' adjective; that is one that applies to individuals only in conjunction with
another description.

P. T. Geach, 'Good and Evil', Analysis, 17 (1956), 33-42.

Geach thus drew a distinction between predicates such as 'large', 'genuine', and
'good' on the one hand and 'red' on the other. The former, but not the latter, need, he
said, to be attached to an individual via a certain concept:something that is to be
called 'large' can only be judged large in a particular class of things; and a genuine
van Meegeren is not a genuine Vermeer. Similarly, a book that is a good soporific
may or may not be a good book; the evaluation is of an individual, but by criteria that
are given by the concept under which it is being judged. Geach insisted that when we
call an action 'good' the word is being used attributively, and so not like the word 'red'
in 'This book is red'. In evaluating, we are thus not predicating a quality in the way
that G. E. Moore's writings suggested when he took 'X is good' as the primary form of
a 'judgement of value'. Nor, says Geach, does moral judgement say that something
or other is 'a good thing' or that 'It is good that' something or other is the case; he is
suggesting that without further explanation such utterances have uncertain sense.
This seems to me to be both right and important. I shall therefore take our question to
be whether just action is good action, not whether justice is good, whether it is good
that people are just, or whether a state of affairs in which there is more justice is so
far forth better than one in which there is less. (Whatever, if anything, any of that
might mean.)

end p.162

What difference will it make if we pay attention to 'good' as an attributive adjective in
something like Geach's sense, and think of moral judgement as having to do with
good action? At once, we see it as belonging, at least superficially, with a set of
evaluations whose correctness or incorrectness is to be judged by more or less
determinate criteria. Already we seem to be far away from the 'boos' and 'hurrays',
and even the feelings and attitudes referred to in expressive accounts of judgements
of good and bad. No one thinks that calling a knife a good knife, a farmer a good
farmer, a speech a good speech, a root a good root, necessarily expresses or even
involves an attitude or feeling towards it. And even a description such as 'good for my
purposes' has to be judged, objectively, by reference to the purposes that I have.
This is not, however, the point that I want to stress here. It is rather that even the
most adventitious list of attributive uses of 'good' will contain examples differing from
each other in striking respects. Thus the last example contained an explicit reference
to the purposes of a particular person, not necessarily wanting what others want;
whereas an object's being 'a good knife' (tout court) must be judged by reference to
cutting; while a root's being a good root depends on something to do with

nourishment or anchorage for a plant; and a farmer is a good farmer only if he looks
after his soil, his animals, or his crops.

See Philippa Foot, 'Goodness and Choice', Proceedings of the Aristotelian Society, suppl. vol. 35

(

1961

), 45-60; repr. in Foot, Virtues and Vices (Oxford: Blackwell, 1978; Oxford:Oxford University

Press, 2002).

Now suppose, for the sake of argument, that the judgement of an action as a good
action belongs somewhere in the extension of such a list. Where would it come? It
would come, I shall argue, in a range of attributions of goodness that could be called
'autonomously species-dependent'. So let us see what this might mean. It concerns,
as the words suggest, evaluations of individual living things, and of their
characteristics and operations; and this without reference to interests that we may
have in them.

I mean by this to exclude, e.g., comfort to a rider in the evaluation of the way a horse moves.

This is what the word 'autonomous' is doing in the description of the mode of
evaluation. How species-dependence comes

end p.163

in must be the subject of further discussion, which may begin with the faculties of
animals, and with certain things the animals do.
Take the operation of sight or hearing for instance, or the power of locomotion. An
individual animal which is well or badly endowed in respect of these things is fitted to
operate correspondingly badly or well.

Whether it actually does so depends, of course, partly on external circumstances and on other things

about itself.

But it is obvious that it is only by reference to its species that we can so evaluate an
individual. It is only as the endowment of a such and such that the perception or
other capacity of any individual can be judged. Nor is it just that in respect of sight
moles are not to be judged by the standards of hawks; as if there were one thing,
'good sight', which different kinds of animals could be expected to have in different
degrees. The example of owls puts us on to the fact that it is the way of life of the
species that determines what the better or worseness of sight consists in for an
individual. And it is, of course, this rather than the circumstance it itself happens to be
in. For an individual put into a special environment, like a wild animal in a zoo, may
have what is no less a defect because it happens to fit its special life. Expressions
such as 'good sight' have reference to a species in their meaning. If we want to talk
about the kind of sight that happened to be advantageous to a particular individual in
a special environment, or even in its own, we can always do so; but then we are
talking not of good sight but of something else.
Thus, where excellences and defects of animals are concerned, it is by reference to
the life of the species that the evaluation is made. But what does 'life of the species'
mean here? It does not include everything characteristic of members of the species,
but only that which has some function in that life. There is, as far as I know, no
distinction to be made (in species-dependent goodness) between leaves that rustle
badly and those that rustle well.
This may begin to give some idea of what is meant by saying that autonomous
species-dependent goodnesses and defects derived from 'the life of the species'. But
how is that to be worked out? The key notion is, I believe, the concept of need
appearing in various constructions; as when we say what a plant or animal of a
certain species needs to have,

end p.164

both in itself and in its environment, and what it itself, and its operative features, such
as roots, leaves, hearts, and lungs, need to do. Good eyes are the eyes a particular

kind of creature needs; good roots the ones a particular kind of plant needs; and so
on. And since the idea is that the organisms cannot flourish without the parts,
features, organs, and operations that things of this kind need, the notion of need is
roughly the one described, most helpfully, by David Wiggins as 'absolute need'.

He is contrasting this with the 'relative need' which would be implied if someone said that he needed

money for an expensive suit. David Wiggins, 'Claims of Need', in Needs, Values, Wiggins, Truth
(Oxford: Blackwell, 1987).

Wiggins defines this as the notion of need that depends on a conditional to the effect
that something or other is necessary for the avoidance of harm. This is not exactly
what I want here because an organism can be defective in lacking something
necessary for reproduction, as an animal is if it loses the mating instinct, and perhaps
this does not exactly count as harm, unless a species itself can be said to be harmed
in this way. There is, however, a more important difference between what Wiggins
has (and needs) for his purposes and what I need for mine. I am interested in
species-dependent goodness, and therefore in what is needed (absolute need) by a
member of a particular species as such. He, on the other hand, is interested in claims
of need, and therefore as much with individuals as with species. An example which
shows the difference is that a fly in a fly-bottle, which will starve if it does not find the
exit, has an absolute need to do this on Wiggins's definition of absolute need. But as
fly-bottles are not part of the natural habitat of flies, and it is no part of the survival
mechanism of flies to distinguish direct and refracted light, this is not, in my sense,
one of the things that 'flies need to do'. No doubt many tricky questions can be raised
here about adaptation to new environments, but this introduces a dynamism into the
model: a dynamism from which notions such as good eyes and good roots must
abstract.
Autonomous species-dependent goodness is therefore intertwined with a particular
notion of need; needs being based, at least in the case of plants and animals, not
only on what is necessary for the health and survival of individual members of a
species but also for the continuance of the species itself. A further distinction is to be
made which depends on

end p.165

the answer to the question 'Where does the harm occur if the need is not met?' For
while a defect of sight, or hearing, or instinct, most often rebounds on the one who
has it—whom we may call the subject of the evaluation—this is not always so. If a
gull cannot distinguish its own offspring's cry in the cacophony of the colony it will be
the chick that dies; and the cubs die, not the lioness, if she fails to teach them to
hunt. There are many such examples. One thinks, for instance, of members of the
species of dancing bees, who dance a complex dance that directs other bees to a
source of nectar. Or again of any species of stinging bee. It is necessary that bees
have stings (they need to sting): so that predators learn to fear bees, and for the
protection of the hive. Bees need to sting. So does this individual bee need to sting,
in spite of the fact that it itself will die if it does? It does not matter exactly how we
choose to speak at this point. What does matter is that the evaluation of an
individual—of its features and operations—may depend on harm, public or private,
which threatens others rather than itself.
Let us turn now to human beings. Are their excellences and defects likewise species-
dependent, despite the fact that there is so much greater diversity in the way they
live? First of all, what about such things as defects of sight or hearing with which we
opened this discussion? It is important, of course, that human beings can sometimes
choose their way of life, and can deliberately modify their environment. What they
need in the way, e.g., of sight may depend somewhat on chosen surroundings.

Moreover, for human beings there are greater diversities of lives within a given
habitat, as for instance in a single city, and this will have to be taken into account.
Nevertheless, a very great deal is in common to all human beings. None of us need
the eyesight of a gannet; not having a gannet's diving capacity, and having other
ways of catching fish. But we all need sight that allows us to recognize faces; and
ears to hear, tongues to sound, and brains to process, speech. To be sure, some
particular professions need specialized capacities, as a scholar needs a specialized
memory. But there are some things quite apart from language that human beings as
such need to be able to remember, like the distinctive physiognomy of friend or foe.
One would want to stress, however, that a great change has come over

end p.166

the conceptual scene with the move from animals (other animals) to humans, and
that we shall have to look afresh at the necessities standing behind the evaluation in
their case. To begin with, we now have to deal with the much larger number of
harms, and here also of deprivations, that can be suffered: deprivations that belong
to the world of imagination and understanding for instance. On the most simple level,
and leaving aside such things as art and science, one notices that a human being
who is unable to follow a tune or a dance is deprived, and therefore does not have
what a human being needs for the good things that human beings enjoy. Although
the notion of deprivation is elastic and difficult, it is right to speak of such human
beings as poorly endowed. But it depends again on human life, on the human
capacity to find things interesting, amazing, and marvellous that we understand the
idea that someone is deprived if he does not have many things that are not
necessary for life or even for health. Flourishing, for human beings, encompasses the
enjoyment of many good things. Anyone is deprived if he does not have some
measure of this enjoyment; and if the deprivation is due to some incapacity in him
rather than to external factors then this is one source of adverse evaluation. I hope
that it will be enough for the purpose of this paper to follow the ordinary use of
'deprivation' and 'need': the one we all understand when it is said that human beings
need

affection, or that lack of affection is itself a deprivation, as hardship is not.

Much of what has to be said here will depend on a feature of the operation of human
beings to which Aquinas drew attention, when he said that while animals perceive
things that are good and go for them, human beings go for what they see as good.

Thomas Aquinas, Summa Theologica, 1a 2ae, q. 1 a. 2 (First Part of the Second Part, question 1,

article 4).

This idea can seem extremely puzzling if one thinks that seeing something as good
ought to be like seeing it with a halo. In fact the truth, and great importance, of what
Aquinas said comes out in the recognition of the whole conceptual structure
belonging to the idea of human action and desire; as for instance in the fact that they
can say why they want something, can conduct practical reasoning, and defend a
choice. It is also important for the question of

end p.167

absolute needs of human beings that there is second-order evil in human life,
meaning for instance the misery that comes from the consciousness of being
disregarded, lonely, or oppressed.
It seems clear that in spite of the greater range of considerations that are relevant to
the determination of absolute need in human beings, the evaluation of such things as
locomotion, sight, and memory, has the same conceptual structure in animals and
men. There is much more flexibility in our judgement, and variation in the
background, but nevertheless physical and mental health, good sight, etc. are judged
here in a species-dependent way. And the tenor of this paper is to suggest that the

same is true when the subject is the goodness or badness of human action. If it is so,
and if, moreover, action is evaluated in relation to human needs, as the things bees
do are evaluated by what bees need, this will be highly relevant to the status of
justice as a virtue. Is the behaviour of an unjust person defective, we should ask, for
the same reason, broadly speaking, as is the behaviour of a free-riding non-dancing
dancing bee, or a lioness who does not teach her cubs to hunt?
On the pattern of the evaluation of other operations in animals and in men we now
need to point to the good served by justice. But surely we can do that? For the
teaching and observing of rules of justice is as necessary a part of the life of human
beings as hunting together in packs with a leader is a necessary part of the life of
wolves, or dancing part of the life of the dancing bee. As Elizabeth Anscombe has
said about one aspect of justice—the keeping of promises—a great deal of human
good hangs on the possibility that one man can bind the will of another by getting him
to promise to do something. As she says, the institution of promise-making and -
keeping is 'an instrument whose use is part and parcel of an enormous amount of
human activity and hence of human good; of the supplying both of human needs and
of human wants so far as the satisfaction of these are compossible. . . . It is scarcely
possible to live in a society without encountering it and even being involved in it.'

G. E. M. Anscombe, 'On Promising and its Justice', in Anscombe, Collected Philosophical Papers

(Minneapolis: University of Minnesota Press, 1981), iii. 18.

Why then will it seem, to most philosophers, impossible to infer from this kind of
consideration, that a human being who refuses to count 'I

end p.168

promised to do A' as in itself a good and rationally compelling (though not of course
always overriding) reason to do A, is defective precisely in practical rationality? Why
will the analogy with defective behaviour in animals seem so ridiculous to many
moral philosophers?
First of all this may be because the analogy seems to give us a picture of morality as
blind adherence to some norm for a species to which we happen to belong. But to
think this is a mistake. For while in a wolf, or a dancing bee, appropriate hunting or
dancing (as opposed to free-riding) is all that is needed for them to be behaving well
in this respect, mere conformity to rules would indeed be blind obedience in one of
us. To advert again to what Aquinas said, human beings not only go for what is good
but for what they see as good. So it is not surprising if practical rationality requires
the understanding of reasons for acting; which is why we criticize some people's
following of social norms as 'merely conventional' behaviour. It may be objected that
most people do not understand why they should act according to justice, or to other
virtues. But this is not true. For this understanding is not something acquired only by
study and debate. On the contrary it is just that which is expressed by any one of us
when we say, e.g., 'How would we get on if people did not keep their promises?' or
'What if no one helped anyone; where would we be then?' This type of answer can
readily be given to the question 'Why should one act justly?'; that is 'Why does one
have reason to do so?'By contrast, similar answers could not be given, or not
successfully defended, if the question were about the rationality of obeying just any
old rule, as, e.g., a duelling rule or a merely snobbish rule of etiquette. It was
because I myself failed to see that this was crucial to the discussion of reasons for
action that I made such a mess of things in an article called 'Morality as a System of
Hypothetical Imperatives'.

Philippa Foot, 'Morality as a System of Hypothetical Imperatives', Philosophical Review, 81/3 (

305-16.

Still more or less in the grip of a desire-based theory of practical rationality I looked
everywhere but where I now think it is right to look for the difference, in the matter of
practical rationality, between one who follows any old rule and one who follows
morality.
It may be objected that the kind of thing that people naturally say

end p.169

about justice or charity does not draw on the idea of the human species which plays
so prominent a part in the argument of the present paper. But this is not true. I have
already quoted Elizabeth Anscombe on the need that human beings have for binding
each other's wills in an enormous number of their enterprises. The reason why we
need mutual aid is also easy to see. If people did not get ill or in other trouble some
at one time and some at another, and if they were not able to give aid without
themselves falling into the same trouble, then there might be no good reason to
consider benevolence a virtue and to perform kindly action just because of that. I
remember that this thought about how different things would be for a race of beings
who could only exchange trouble among themselves, so that one person could be
got out of a pit only if someone else fell in, was one of the ideas that first put me on to
the possibilities of a species-dependent account of virtue.
A second objection may, however, seem to be more powerful. It has to do with a
shameless individual, concerned perhaps only about his own power, wealth, and
comfort, who says that nothing said here has given him any motive for acting justly
when such action is unlikely to pay off in his terms. But my argument was not
supposed to show that such a man would have a motive for just action, if 'motive'
means, e.g., what the detective looks for after a crime. What I was arguing was, that
since it is part of practical rationality every human being has reason to be charitable
and just. If our shameless immoralist says that he has no reason so to act he is
simply begging the question. He has, to be sure, nothing that he recognizes as a
reason; but that, if my argument is correct, is because he fails to recognize the truth.
Any idea that if he does not have a motive he cannot act justly is of course ridiculous.
No one says this unless he is influenced by bad philosophy, perhaps of the kind that
interprets 'motive' half in the ordinary way and half as 'whatever it is that moves us to
action', thinking that there must always be such a 'moving force'. Considerations such
as 'I promised' or 'It is his property' do very generally move people to action. If they
did not they could not play the part in human life which they do. By hypothesis they
do not move the totally selfish, shameless man, and perhaps there is no way in which
we can touch his life. But it is an advantage, not a disadvantage, in an account of

end p.170

reasons for action, that it leaves room for such a person. To show that there is
reason for him, as for everyone else, to do what justice demands, we do not have to
show how can be got to act.
This concludes what I want to say here about the rationality of justice; which I believe
can be applied, mutatis mutandis to other virtues which have to do with the good of
others rather than oneself, as, e.g., charity and parental devotion. It is time, therefore,
to confront the theories of practical rationality which, I suggested, were wrongly seen
as preconceived structures into which we must try to fit the rationality of moral action.
We have to deal with these theories—or at least with the examples of irrationality on
which they relied—if the concept of practical rationality is not to break apart. We have
to take account of the irrationality of self-destructiveness for instance, or with selling
one's birthright for a mess of potage. And I think here of a burglar I once read of who
sat down to watch television in the house he was burgling, and got caught. Here too
there were good reasons for acting otherwise, whether the agent recognized them or

not. Through his lack of concern for consequences he was so far forth deficient in
practical rationality, so that he acted badly not only in thieving but also in this respect.
Do we then have two different kinds of rationality, with a theory for each; one to fit the
examples just given and another to fit the rationality of acting justly or according to
other virtues? Not at all. For if I am not mistaken, the basic ground of the rationality is
in both cases the same.
Why, after all, do most of us so readily accept that a rational person takes account of
his own good, weighing advantages and disadvantages of action by their effect on
his future, even his distant future? To do so is a possibility for human beings as it is
not, or not in the same way, for animals, because language allows us to throw our
thoughts into a hypothetical future as they cannot. Moreover, human beings need to
do this. It is part of the way they survive and flourish. Unless others looked after him,
the life of one who never came to do it would be nasty, brutish, and short. Moreover,
it is a feature of the way we operate—one so general that we hardly notice it—that
each one of us takes special thought for his or her own life. Here some variation is
possible. We hire doctors to tell us what to do for our future health, and might give
them more power to act

end p.171

for us than we do. But it would obviously be ridiculous for everyone to act for
everyone: no one can see more than a very little on so vast a scale. A sort of 'buddy
system' might be suggested, by which each person looked after one other, and was
looked after by one other in his turn. But no one can plan for another human being as
he can plan for himself, if only because intentions form so large a part of plans. Our
behaviour would therefore be defective if we had no special care for ourselves.
There are many facets of practical rationality which seem to be based in the same
kind of necessities of human life. There is, for instance, the properly careful
consideration of means to ends, and the willingness to take the best means even
when they are toilsome or alarming, or when they involve the denial of an urgent
wish. To fail in this can be in itself a fault, as it was with the TV burglar. This does not
of course imply that ends are neutral from the point of view of practical rationality.

my argument is correct the burglar had two faults, both failures of practical rationality:
one because he failed to see (or to act on) the fact that there is reason not to steal,
and another because he failed to see (or to act on) the fact that there is reason not to
risk imprisonment for half an hour's TV.
It would have been natural to end the preceding paragraph by saying that there was
a more compelling reason for leaving the house quickly than there was for sitting
down in front of the television. And it is time now to say more about the relative
strength of reasons, which I have hardly touched on yet. Is this, too, determined,
when determinate, by facts of human life? It seems to me that it is. It is obvious, for
instance, that the fulfilment of promises must take precedence—some precedence—
over the plans and interests of the promiser; otherwise the power to bind the will of
another through promises would have little strength. It is obvious too why it is not a
fault to give one's own serious interest, and the interest of one's family, some
precedence over that of strangers; and obvious why rationality demands that in
general parents put the good of their children before their own. This, one would say,

end p.172

reverting to the concept of absolute need, is what needs to happen for the averting of
harm and provision of the kind of good without which human beings can be counted
as deprived.
Nothing of this implies that such questions of precedence are always easy to answer;
nor does it rule out large areas of indeterminacy. A special problem of precedence

also arises from the distinction of greater and lesser human goods. Some things are
important in human life, while others are less important or trivial; and wisdom, as part
of practical rationality, must take account of this in governing our aims. I cannot even
begin to deal with this topic here, except to notice a conceptual connexion with the
concept of deprivation (as opposed to hardship). It is a reasonable assumption,
however, that the idea of importance must depend on facts about the things that run
deep in human life, however exactly that should be understood.
I would claim that in spite of many unsolved problems, the general picture is clear.
The suggestion is that the notion of practical rationality is correlative to that of the
goodness of action, so far as that consists in the proper following of reasons. What is
'proper' or 'good' here is determined by human life and its necessities, analogously to
the way in which good or (proper) sight or locomotion or memory is determined in
both animals and men. Thus practical rationality includes prudential self-interest, the
weighing of advantages, the adoption of means to the securing of ends, but also
such 'other-regarding' matters as care for offspring, fidelity to contracts, and mutual
aid. One has no reason to say that practical rationality has to do especially with the
agent's own self-interest or his own desires, except when that has rational
precedence, as it does sometimes but certainly not always. Nor is there any point in
keeping bits of the vocabulary to signify practical rationality in one particular aspect.
We could of course do so, as, e.g., by using 'irrational' to mean something like 'acting
against one's own interests and desires', and 'rational' to mean 'not-irrational'. Such
is, I suppose, the intention of those who are ready to speak of 'rational villains'. But
this decision would make akrasia as such not irrational, which it is normally supposed
to be. More-over, there would be only awkward ways of expressing the contrariety to
rationality of breaking promises or neglecting one's children. The

end p.173

suggested usage belongs, I think, with the whole idea of a preconceived, 'desire-
based' concept of practical rationality; which it is the intention of this paper to subvert.
The strategy I adopted was to set aside these received theories of practical rationality
which raised problems about the rationality of moral action, returning to them only at
the end of the paper. I argued that if justice is a virtue it must make action good by
disposing its possessor to goodness in practical rationality; the latter consisting of the
right recognition of reasons, and corresponding action. I then argued that justice is
indeed a virtue, having introduced the idea of 'autonomous species-dependent
goodness' to show how I thought that the argument should go. Finally, I suggested
that the examples of prudential and desire-fulfilling actions, on which the
preconceived theories had been based, were themselves in need of the same kind of
scrutiny, and could in fact be validated in the same way. We were left, therefore, not
with two rival accounts of practical rationality, but with only one.

end p.174

Moral Dilemmas

Revisited

Philippa Foot

I have called this paper 'Moral Dilemmas Revisited' to mark continuity with a paper I
published ten years ago under the title 'Moral Realism and Moral Dilemma'.

Journal of Philosophy, 80/7 (July

1983

), 379-98; ch.

in this volume.

The point of that piece was to challenge an argument against moral realism put out
by Bernard Williams in the 1960s, in two articles: 'Ethical Consistency' and
'Consistency and Realism'.

Proceedings of the Aristotelian Society, suppl, vol. 39 (1965) and 40 (1966). Both articles are

reprinted in Bernard Williams, Problems of the Self (Cambridge: Cambridge University Press, 1973).

Moral realism is not the topic of the present paper. But I want to look again at moral
dilemmas. Ruth Marcus was one of the first to arouse my interest in the subject, and
it seems fitting to offer this paper for her Festschrift. She would, I think, be the last
person to want the affection and respect of her friends to take the form of agreement
rather than challenge; so I hope I shall be offering her something she will like.
Marcus's views on moral dilemmas appeared in an article called 'Moral Dilemmas
and Consistency', where, very controversially, she insisted that it is possible to be in
circumstances in which one will be guilty whatever one does: 'damned if one does a
certain thing and damned if one doesn't' as she says.

'Moral Dilemmas and Consistency', Journal of Philosophy, 77/3 (Mar. 1980), 121-36.

This is the thesis that I want to oppose. I do not agree with her that the feelings of
guilt that people often

end p.175

have when they have had to make a difficult moral choice imply that they are guilty;
or indeed that we can even understand the possibility of inescapable guilt, unless we
believe for instance in a cruel God who damns us for disobeying either of two laws on
occasions where both cannot be obeyed.
With these doubts I shall also once more be querying a thesis put out by Bernard
Williams, who, like Marcus, thinks that feelings of guilt should be taken very
seriously, and who has erected upon such feelings a doctrine of the possibility of an
agent faced with inevitable wrong action. How, asks Williams, can we rule this out?
What facts about the world or about an agent could do so?
There is a substantial and interesting question:'What would have to be true of the
world and of an agent that it should be impossible for him to be in a situation where
whatever he did was wrong?' I doubt in fact that there is anything that could produce
such a guarantee short of the existence of a rather interventionist God, or else the
total reduction of moral life to rules of efficient behaviour—two extremes which
precisely leave out the actual location of moral experience. But it is at any rate a real
question, and it would not be a real question if the correct thing to say were that
nothing has to be true of the agent or of the world for this to be so, because it is
guaranteed by the logic of moral expressions.

Bernard Williams, 'Conflicts of values', in Williams, Moral Luck (Cambridge:Cambridge University

Press, 1981).

In the end I shall focus on Williams's view, which seems more basic than that of
Marcus, in that her thesis requires the truth of his. For she wants to say not just that
feelings of guilt are natural and even creditable after a hard choice, but also that the
subject who has them may actually be guilty whichever horn of a dilemma he
chooses to grasp. And for this it is not enough that he can't get away from the idea
(and so from the feeling) that he has done something wrong; it must be true that he
has.
Marcus's position is quite extended; since she thinks that the agent's choice must
leave him just a little guilty even in the case of a moral dilemma in which 'priorities
are clear and overriding'. She says that in such cases the guilt is 'appropriately
small'.

Marcus, 'Moral Dilemmas and Consistency', 130.

But why does she think that there

end p.176

is any guilt at all, when, for example, someone justifiably breaks a promise in order to
fulfil a more pressing obligation? What she is really interested in is, it seems, not guilt

but rather 'remainder': she wants to deny that an obligation that is overridden simply
disappears. So we should look at the doctrine of 'remainders' with respect to moral
dilemmas, found not only in Marcus but also in Williams and others, and see what, if
anything, it has to do with guilt.
A moral dilemma can be characterized, first of all, in terms of evidence. Seen in this
way, a moral dilemma (which, as understood here, need not be such as to put
anyone in his senses into any uncertainty as to what to do) will have to be a special
case of the dilemma that exists wherever there is evidence for and evidence against
a certain conclusion. What is special is that the conclusion is about what an agent
ought to do. The relevant "ought" signifies what Williams has felicitously called 'the
practical ought', which he has also called the ought of advice. Moreover, it speaks of
what should be done not from some special point of view, but rather all things
considered, and gives a verdict, where there is a verdict to be given, about which of
two conflicting principles overrides the other. In the present context it may be thought
of as the moral "ought", so long as moral judgements can be assumed both valid and
overriding. One should note, however, than in some other discussions this would be
in question, as when Thrasymachus, giving advice, would say that anyone strong
enough to get away with injustice should try to do that.
Let us represent this practical, all-things-considered, ought with the formula OXA for
"X ought to do A". And now let us consider the evidence or grounds for such a
judgement. These grounds are expressed in a second class of propositions, stating,
inter alia

, the existences of rules, promises, engagements, and considerations about

good and harm to oneself and others. There is also a third class of propositions that
assert an evidential relation between such things and the 'practical ought' (i. e. which
assert their status as grounds). So, still, for the moment, speaking of evidence, we
say for example that where X has promised to do A (call this consideration C) there is
reason to believe that OXA; or "Prima facie, on C, OXA".

end p.177

What is confusing here is that there is an ambiguity in "ought"which allows a
proposition about what an agent ought to do to be either of the type to which we have
so far assigned it, in which it stands for Williams's 'practical ought'; but also, along
with assertions about the existence of obligations, promises, rules, benefits, and so
forth, to appear in the propositions that serve as grounds. The term "obligation" is
even more ambiguous; it being quite natural to use this word not only for the all-
things-considered verdict in moral contexts but also to mark, for example, contracts,
and claims engendered by special relationships. John Searle has suggested that it
will be useful in this discussion to restrict "obligation" to this use, while always
understanding "ought" to signify the 'practical ought'.

John Searle, 'Prima Facie Obligations', in Z. van Straaten (ed.), Philosophical Subjects: Essays

Presented to P. F. Strawson

(Oxford: Clarendon Press, 1980).

Alternatively we could flag the "ought"which speaks of a ground for the practical
ought

in representing it, as well as Searle's obligation as O′; when we shall have

OXA, O′XA, and Pf. OXA/O′XA.
What is the connection of all this with reasons for action? We notice, first, that there
is an evidential relation parallel to the ones we have been considering, but this time
between there being a reason for X to do A and there being best or conclusive
reason for the same thing. So we have R XA (best reason), R′XA, and Pf. R XA on
R′XA. Moreover, since O represents the practical ought of advice, OXA implies that
there is, all things considered, best reason for X to do A. It is evident, however, that
in making this explicit connection between OXA propositions and reasons for action,

and in characterizing O′XA propositions as their ground we have tacitly restricted the
O′XA range to propositions that themselves give reasons for acting, or at least give
evidence for the existence of such reasons.

Derek Parfit has pointed out to me that the testimony of others can be evidence of reasons for acting

without itself being such a reason.

It is open to anyone to question whether such things as promises, family
connections, or considerations about the good of oneself or others, really do give
reasons for action and therefore evidence for propositions of the OXA form; but this
is not my problem here. I shall proceed on the assumption that it has been solved,
whether by showing that rationality is not a morally neutral notion, or by succeeding
in joining a neutral concept of rationality to a standard, or fairly standard, moral code.

end p.178

I shall therefore suppose, for the purpose of this discussion, that there are true
propositions of this kind, represented in the formula OXA, and naturally read "X ought
to do A" (or "should do A") where this implies that there are conclusive or best
reasons for X to do A. I shall represent an alternative to A by the letter "B" with
subscript " A"; meaning to cover both the cases in which B is an action specified by a
description other than A and that in which this is its only specification. Thus I can
speak alternatively of the dilemma about whether to do or not do A, or about whether
to do A or B. In the latter case A and B will be consistent in the sense that at least in
some possible world they could both be performed by a single agent on a single
occasion, although they are such that in our world they cannot, in the circumstances,
both be performed.
On this supposition we have, then, a conflict of reasons for action; in the sense that
there are actions A and B

and one reason at least for the doing of A and another

for the doing of B

. And with the explicit introduction of reasons for action, as

opposed to evidence for a proposition about what ought to be done, we have
reached the place at which many philosophers have argued, rightly I think, that there
is a special "remainder"when a conflict of reasons has been settled either in the way
it is settled when one reason or set of reasons has been found stronger than the
other, or in the way it is settled when the dilemma has seemed insoluble and the
agent has simply decided how to act: the suggestion being that this kind of remainder
is not found in cases of evidence for a merely theoretical conclusion.
Considering "remainder" I shall think first, for the sake of simplicity, about resolvable
moral dilemmas; that is, moral dilemmas where the evidence for OXA is stronger
than the evidence for OXB

; where an obligation to do A would override the

obligation to do B

, and there is stronger reason for doing the first action than the

second. The familiar old example is that of someone who has promised to be at a
certain place to meet a friend but who has instead to drive an accident victim to
hospital. Obviously he or she ought, all things considered, to do the driving rather
than the meeting, and the question is what "remains" from the obligations, reasons,
and so on, on the other side.
First of all, let us consider "linguistic remainder", that is what can we say in such a
case if employing the same terms as those appearing in the

end p.179

O′XA or grounding proposition. With some considerations this is simple. A promise is
not annulled when its practical force is overridden; and when one has "double-
booked"engagements, the one that it is right to break is not any the less an
engagement for that. These bits of language clearly work this way, but that is not so
obvious in the case of obligations, and certainly not in that of reasons for action.
Perhaps it is right for obligation, for it seems to make sense to say that one has an

obligation that one cannot fulfil, and on that principle perhaps one can say that one
has an obligation to do B

even after the doing of A has made this impossible, so

long as the time for doing it has not passed. But this does not seem to be true of
reasons for action. For while it is right on occasion of conflicting reasons to say that
"there is a reason"for doing each action (though not, of course for doing both)
whatever the relative strength of the inescapable reasons may be, it does not follow
that one still has a reason to do A when B

has been done. This is partly because

the other action is no longer possible and it is not clear that there can be a reason for
an agent to do what he cannot do, and partly because the doing of one action may
have destroyed the desirability of doing the other. Williams has pointed this out, with
an example having to do with the desirability of courting, or marrying, either of two
women but not both. And this is, of course, an extra reason, beyond the principle that
ought implies can

for dropping what he has called the principle of agglomeration in

this area.

Williams, 'Ethical Consistency', in Problems of the Self, 180. See also Marcus, 'Moral Dilemmas and

Consistency', 134.

O′XA implies that X can do A, and O′XB

implies that X can do B

;but as A and B

, are incompatible we cannot factor out to get O′ (A and B

). And similarly there

may be reason for doing each but no reason for doing both. So when one of the
actions has been done there may no longer be any reason for doing the other; not, or
not only, because one can have reason only for doing what one can do (if that is true)
but also because the doing of either action (the better or the worse) may always have
destroyed the reason for doing the other.
For some of these "remainders" one may, therefore, have to take time into
consideration when asking what effect being overridden in the practical or best
reason stakes can have on the original grounding propositions.

end p.180

In some cases they can still be asserted; with a change of tense only at some time
explicitly or implicitly referred to in the description of the action; but in other cases the
doing of the alternative action may mandate a change of tense.
None of this is, I think, very interesting compared with other reasons that have been
given for the presence of "remainder" even in cases of resolvable moral dilemma. For
instance, Susan Hurley believes that the possibility of akrasia shows that reasons for
acting which are overridden do not ipso facto disappear; on the ground that in akrasia
someone acts on a reason that they know to be other than the best.

Susan Hurley, Natural Reasons (Oxford: Oxford University Press, 1989), ch.

, sect. 3.

This seems to me interesting but not quite right, on account of a distinction that I
think must be made between what one might call (a) explanatory and (b) validating
reasons, and the fact that an explanatory reason is (if one is indeed needed) enough
for the possibility of akrasia. Consider, for instance, someone who does something
for the sadistic pleasure that it gives him. Does he have a reason for what he does?
In one sense he does, because there is a particular kind of explanation of the action:
in terms of what Anscombe has called a ' desirability characterization'.

G. E. M. Anscombe, Intention (Oxford: Blackwell, 1957), sect. 37.

If he says this was the reason why he did it, this gives his action a particular kind of
intelligibility, and we could speak of"his reason"here. But we ought also to say,
surely, that the fact that it will give the agent sadistic pleasure is no reason—not any
reason—for doing it, as he himself may admit. Indeed, as Gavin Lawrence has
pointed out to me, the fact that he will get pleasure from doing something may rather

seem a reason against doing something than even the slightest, most overridable,
reason on the other side, and he himself may even believe that.
With the existence of a reason for rather than against doing an action we are in the
area of validating, or as I should like to say (in non-Davidsonian terminology)
"rationalizing reasons": that is, reasons having to do with the rationality of the action,
not merely with its explanation.

For Donald Davidson's terminology, see 'Actions, Reasons, and Causes', in Davidson, Actions and

Events

(Oxford: Clarendon Press, 1980), 3.

end p.181

I do not know whether this distinction will be readily accepted. If not this may be
because the question of rationality can come in in a derivative way once an agent
has determined on the pursuit of any end, however bad that end may be. There can
be calculation and therefore practical reasoning about how to achieve any end, child
abuse for instance, and a lack of rationality in relation to this end.

I once read of a burglar who sat down to watch television in the house he was robbing, and got

caught. He suffered from a lack of rationality that was additional to any that belonged to the project as
a whole.

But this does not mean that the agent even thinks there is reason to pursue the end.
If he calculates well and follows where the calculation leads him he does indeed
avoid one failure in rationality; but there could nevertheless be no reason to pursue
the end he is pursuing, and strong reason against it. His desires have no doubt
played a part in determining what he does, and he may have held a "desirability
characterization"in his mind when he was taking the decision about what he would
do. He lets it affect him in a way that is familiar to us all, particularly when we want to
do something we know we ought not to do. And the practical ought may also
determine our action. So three different stories are told here. There is the story about
influences, in which the all-things-considered ought is (regrettably) only one
competitor among others. Then there is the story about the weighing of reasons, in
which the practical ought might seem to, but does not, have a speaking role; since it
is not a pro among the cons. And finally there is the answer to the question "What
ought I to do?"in which no character except the practical ought is on stage.
The distinction between rationalizing and explanatory reasons is important, but I want
it at the moment only in order to show why I think that Hurley is not quite right in
saying that the possibility of akrasia shows that reasons for acting remain even when
they are overridden by stronger reasons on the other side. For it could be (though it
no doubt isn't) the case that all akratic actions were of the sadistic pleasure or child
abuser's type; where "his reason"means only the "desirability characterization" that
makes what he did intelligible in a particular way. I agree with Hurley and others who
have said that there may be proper (rationalizing) reasons for doing both of the
incompatible actions A and B, and of course in many akratic actions there are these.
This seems to me

end p.182

to be enshrined in the very ordinary notion of the pros and cons of a given course of
action. For it isn't right to say 'the supposed pros and cons' with the idea that either
the pros or the cons are going to disappear if the conflict can be resolved.
It makes good sense, therefore, to say that there are, for a given agent in given
circumstances, reasons to do A and reasons to do B

, even when one set of

reasons is stronger than the other. But it is not easy to see what more should be said
about "remainder" for moral dilemma in general. Those who stress this remainder are
apt to talk about the moral requirement for apology and restitution to someone whose
interests have been set aside. But this cannot always be in question, since a moral

dilemma may involve only the agent's own life. Moreover, even where others are
involved, as in the case of promises, it does not seem obvious that we owe anything,
except an explanation, to another when we have justifiably broken a promise to him.
Nor need our broken promise even have caused him any inconvenience, as we can
see by changing the old example to make the circumstances of the two sides
symmetrical. Now there are two accidents, and we meet, each with our accident
victim, at the door of Casualty; not even having to explain why we are not where we
promised to be. In this case no one suffers any inconvenience from the broken
promises, and we can suppose that for some reason things turn out better for each of
us than if the other's promise had been kept. (Perhaps a bomb went off in the
building in which we were to meet.)
All of this shows that the supposed demand for apology and restitution will not serve
to fill out the idea of "remainder"for moral dilemma as such. Especially in the case of
overridden promises, there may be nothing to it except that we would have kept our
promise if we could have done, and that it would have been incumbent on us to look
around for a way of removing the dilemma if we could; perhaps asking another driver
at the scene of an accident if he or she were freer than ourselves.
Nevertheless, Williams was surely on to something important when he wrote of
'moral cost' and 'moral loss' in connection with the choice of one horn rather than the
other in a moral dilemma.

See e.g. 'Conflicts of Values', in Williams, Moral Luck, 76.

This does seem appropriate at least for many cases, where it is not possible to
winnow

end p.183

out the loss of good as in the previous example. Isaiah Berlin has persuaded us to
take very seriously the idea of an inevitable loss of value—the necessity of sacrificing
one value to another—both in political decisions and in an individual's choice of
lives.

See Isaiah Berlin, 'Political Ideas in the Twentieth Century', sect. iii, in Berlin, Four Essays on Liberty

(Oxford: Oxford University Press, 1969) and the introduction to that volume.

Berlin writes very naturally in this context about what we might call human goods
such as life, health, liberty, and knowledge. And one notices that this natural way of
talking can form a basis for assertions to the effect that it is good (or bad) that such
and such happened, or is the case. Saul Kripke once gave me, when I was querying
the objective use of such locutions, the example of the preservation and discovery of
the Dead Sea Scrolls. This preservation and discovery surely was an excellent and
important thing, and in saying so we do not simply reflect a personal wish or concern,
any more than we would in speaking of the preservation and discovery of some great
piece of music or of a great scientific discovery. A whole area of impersonal
utterance belongs here; including all that we spontaneously say about "great" or
"marvellous" things, about what is "deep", and about "the things that matter".

We understand very well what it means when someone says on his deathbed that he wasted his life

on trivial things: on 'things that didn't matter'. Also when Berlin writes, again quite naturally, of 'the best
things in life'.

The idea of value will come better into focus if we explore such utterances rather
than the bare "good"and "bad", which, on their own, suggest Moorean intuitions of
goodness and badness in happenings or states of affairs.
There is, therefore, no doubt but that Williams is right in thinking that there is
something we might call "moral loss" or "moral cost", at least in the cases of moral
dilemma that Berlin has stressed. For we may have to sacrifice one thing that we
value for the sake of another, such as family for achievement, or, as is only too

frequent in times of war, great architectural monuments for lives. And indeed most
moral dilemmas concern what might intelligibly be called "human goods"such as
health, life, liberty, and knowledge.
It is not so clear, however, that this will go any way to vindicating Williams's idea of
inescapable wrong action and therefore (possibly)

end p.184

Marcus's idea of inescapable guilt; because it does not seem that the
"badness"comes in the right place. To see that one might consider, as an analogy,
dilemma in the matter of musical performance. What does it mean to say that a
musician played in the wrong way? There has got to be something bad about his
performance; but obviously it is not enough that it was bad that he played as he did,
since we might say that when a passionate pianist brings down the ceiling with his
rendering of Rachmaninov. Nor, more relevantly, is it enough that there is some
value, and moreover a musical value, sacrificed in his playing; as, for example, pace
for expressiveness; since this might be the very best way to play the piece. And it
makes no difference if we suppose an insoluble musical dilemma, where there is
equal or incommensurable loss in each of two incompatible renderings; and one has,
as Berlin has put it, to 'plump' for one or the other.
Let us turn, therefore, to another way in which Williams has tried to fill out the
concept of remainder; that is, with the idea of moral regret.

See e.g. Williams, 'Ethical Consistency', in Problems of the Self, 172 ff.

At first sight it may seem that this is simply the psychological counterpart of moral
loss

, and so something already recognized. But investigation shows that this is not

right. It is true that in a few rather formal contexts regret may be for anything seen as
bad, as when the bank manager regrets to inform us that we are overdrawn. But
these are somewhat peripheral uses. Normally, regret is spoken of only when a
choice or other action is in question. "Do you regret your divorce?" implies that one
chose the divorce or at least consented to it and now perhaps wishes that one hadn't.
If one is burgled one is sad about having been burgled; but it is the burglar, if anyone,
who regrets the burglary; whereas what one oneself regrets is carelessness in
locking up. Moreover, whereas sadness can be about an action necessary to
implement a choice that one thinks was all things considered for the best, this is not
true of regret. So acting for the best in a moral dilemma, while it can entail sorrow,
and in serious cases even horror, does not, if we have no doubts about the rightness
of the action, make a place for regret. It is when there is no clear answer (and
perhaps no answer at all) to the question of what we should do that we are likely to
waver, so that some days we regret our action and

end p.185

some days we do not. Possibly it is on this account that those who stress
"remainder"are apt to pay otherwise puzzlingly special attention to irresolvable
dilemma or otherwise difficult choices. This special attention is found even in Marcus,
although her thesis is supposed to hold for all moral dilemmas.
For Williams it is essential to give special attention to what he and others have called
'tragic choices', where a high degree of seriousness is combined with uncertainty,
because it is here that he locates the possibility of inescapable wrongness; seeing
this as something that the agent is faced with only in very exceptional cases, as
when Agamemnon must sacrifice his daughter or lose his army. This "wrongness" is
supposed to be something more than the "moral disagreeableness" of dirty hands—
the dirty hands that both Williams and Stuart Hampshire, following Machiavelli, have
described as being hardly avoidable in politics or in war.

See Williams, 'Politics and Moral Character', in Moral Luck, 54-70; and Stuart Hampshire, 'Public

and Private Morality', in Hampshire (ed.), Public and Private Morality (Cambridge: Cambridge
University Press, 1978).

There is no difficulty in understanding this "dirty hands" thesis, since even "dirty
tricks"are sometimes justified, as they surely were when the Resistance fought the
Nazis; and I suppose there can be no dirty tricks without dirty hands. But then those
with dirty hands did not act wrongly, even though Williams seems to think that even
here an innocent victim could complain that he had been wronged. Has he really
been wronged if harmed, even if it was a justified action that harmed him? I should
have thought not, because if the action was justified, then if he does not recognize
that, he is making a mistake. It will no doubt often be difficult to get someone who
has been harmed or "used"to recognize the justifiability in a debatable case. But if he
agrees with us that our action was justified, and even morally mandated, he must
agree that we had a right to do it, and therefore, one would have thought, that he was
not wronged.

Williams himself sometimes says that where an overriding obligation justifies the harm inflicted on

someone he has the right to complain but is not actually wronged. See 'Politics and Moral Character',
in Hampshire, Public and Private Morality, 61 ff.

It is, in any case, in the very special situations—of terrible and insoluble dilemmas—
that we are supposed to find the possibility of

end p.186

inescapable wrongness that Williams insisted on in the passage I quoted at the
beginning of this paper. But even here it is hard to understand what kind of
wrongness

Williams has in mind. It is clear that "wrong" could not there mean what it

usually does. For suppose an insoluble moral dilemma of the kind that, according to
Williams, leaves an agent with the choice between two wrong actions A and B

;

and then a change by which one of the alternatives—say A—has been made even
more ghastly, so that B

is clearly the right thing to do. Then either B

is still

wrong, with a wrongness that is not incompatible with rightness; or else its
"wrongness" did, in some strange way, come from the fact that the old dilemma was
irresolvable; in spite of the fact that genuine irresolvability seems to give us moral
leeway to do either of two actions, rather than to confer a special kind of
"wrongness"on each.
What can it be, then, that makes philosophers feel that they know what the
inescapable kind of wrongness might be? One factor may be the idea of a "moral
requirement"such that there can be one "requirement", as Williams says, on the
agent to do A, and another requirement to do B

Williams, 'Conflicts of Values', in Moral Luck, 74-5.

It is not, however, clear what is meant by "a requirement"here. One wonders whether
there may not be echoes here of the 'divine law' conception which Anscombe has
diagnosed in the contemporary moral philosophy which officially repudiates it?

G. E. M. Anscombe, 'Modern Moral Philosophy', in Anscombe, Collected Philosophical Papers

(Minneapolis: University of Minnesota Press, 1981), iii.

We do, of course, understand the idea of a moral requirement in other contexts, as
we might have said earlier that in cases where someone's claims have been
overridden by a conflicting obligation restitution is 'sometimes required'. But here
"requirement"is explained in terms of a newly generated obligation, and so of
wrongness of the common or garden kind, whereas "requirement" itself needs
explanation if it is to help with the special sense of "wrong" that we are supposed to
understand for cases of "tragic choice". Another contributing factor may be a certain
schizophrenic tendency among moral philosophers who cannot bring themselves to

say that there is any kind of action that is wrong in any circumstances whatsoever,
and who yet feel that certain actions—say torture—cannot ever be justified. For it

end p.187

is exactly the "tragic choice" situation in which they feel they must say that such
actions would be justified, and yet feel that they must always be wrong. To this one
would reply that either the "absolution" of Aristotle, Aquinas, and Anscombe is (as I
believe) correct or it is not. If it is correct, and applicable to torture, then each and
every torturer acts wrongly. Nor should it be supposed that two absolute moral
prohibitions, which will relate to intentional actions, could conflict. For however
terrible the inescapable choices that people have to make, they will never be
between two intentional actions, as, for instance, torturing X and torturing Y, but only
between torturing X and not preventing another from torturing Y, or from bringing
about some other horrible result.

See G. E. M. Anscombe, 'Modern Moral Philosophy', in Anscombe, Collected Philosophical Papers

(Minneapolis: University of Minnesota Press, 1981), 34 ff.

If, on the other hand, absolutism is not correct, then there could at least be
hypothetical situations in which torturing is not wrong, and those who reject
absolutism will have to bite the bullet about this.
However good these explanations may be of why "wrong if you do, wrong if you
don't" has seemed intelligible, they bring us no nearer to actual intelligibility. What
gets in the way is the fact that "wrong" as understood in moral contexts applies to
actions that count against a person's goodness: the goodness spoken of in the
serious, non-ironical, designation of an individual as one of the great and the good.
So if we want to accept 'wrong if you do, wrong if you don't'as an intelligible
possibility, and still keep this negative relation between doing what is wrong and
personal goodness, we have to revamp the latter notion to make that goodness
vulnerable to the "taint" of involvement in a horrifying, humiliating, or tragic situation,
or to the hatred of the gods; and then say that a choice that involves such badness is
"wrong". I do not see how, ahead of having made these or some other shifts of
meaning, we can know how to interpret the idea of a situation in which someone will
necessarily be wrong whatever he does.

end p.188

Does Moral Subjectivism Rest on a

Mistake

Philippa Foot

This paper is about the moral subjectivism that, for the last sixty years or so, has
dominated moral philosophy in England, America, and other countries in which
analytic philosophy is taught. This is the subjectivism—often called 'non-
cognitivism'—that came to the fore with A. J. Ayer, C. L. Stevenson, and Richard
Hare, informed the work of John Mackie and many others, and has lately appeared,
refreshed, in Allan Gibbard's 'expressivist'account of 'normative'language.

A. J. Ayer, Language, Truth and Logic (London: Gollancz, 1936); C. L. Stevenson, Ethics and

Language

(New Haven: Yale University Press, 1945); R. M. Hare, The Language of Morals (Oxford:

Clarendon Press, 1965); J. L. Mackie, Ethics: Inventing Right and Wrong (Harmondsworth: Penguin,
1977);A. Gibbard, Wise Choices, Apt Feelings (Cambridge, Mass.: Harvard University Press, 1990). I
should mention here that although he has not abandoned his attack on what he calls
'Descriptivism'Hare does not want to be called either a subjectivist or a non-cognitivist. See R. M.
Hare, 'Objective Prescriptions', in A. P. Griffiths (ed.), Ethics, Royal Institute of Philosophy Lectures
1993 (Cambridge: Cambridge University Press, 1994). Also, R. M. Hare, 'Off on the Wrong Foot',
Canadian Journal of philosophy

, suppl. vol. 21 (1995).

Simon Blackburn, reviewing Gibbard's Wise Choices, Apt Feelings, has said that he
hopes this book will set the agenda for moral philosophy for the next fifty years. I
myself, for all my admiration for Gibbard, hope that it will not do that. So I should say
why I believe that these non-cognitivist theories—one and all—are based on a
mistake.

end p.189

To identify the common characteristic of the apparently somewhat diverse moral
philosophies that I have just grouped together, and also to do justice to them, it will
be good to start by asking how the whole noncognitivist business began. One finds
its deepest roots in Hume. But more immediately, Ayer and Stevenson's emotivism,
like Hare's prescriptivism, came into being as a result of 'the linguistic turn',
popularized by logical positivism but developing far beyond it. For with 'linguistic
philosophy' came the idea of explaining the singularity of moral judgement in terms of
a special use of language, called 'evaluation' but more akin to exclamation and
command than to anything one would normally mean by that term. With this idea, it
seemed possible, at last, to say clearly what G. E. Moore had meant, or should have
meant, when he insisted that goodness was a special kind of 'non-natural'property.

G. E. Moore, Principia Ethica (Cambridge: Cambridge University Press, 1903).

In the development of emotivism and prescriptivism the idea of a special ('non-
natural') property was replaced by that of a special and essentially practical use of
language. And this, it seemed, was a great discovery. The language of evaluation
was 'emotive'. It expressed a speaker's feelings and attitudes, as well as inducing
similar feelings and attitudes in others. Those who had these 'attitudes''favoured' the
things they called 'good': the idea of an attitude being linked to a tendency to act.
Such also was the doctrine of A. J. Ayer; and a little later R. M. Hare tied 'evaluation'
even more closely to individual action, in his theory of universalized imperatives by
which a speaker exhorted others and, in the acceptance of a first-person imperative,
committed himself to choose what he called 'good'. So 'prescriptivism'—a distinctive
version of the doctrine that I have in my sights—was added to the emotivism with
which it had started out. In an explicit definition of the 'prescriptive' use of language
Hare wrote 'We say something prescriptive if and only if, for some act A, some
situation S and some person P, if P were to assent (orally) to what we say, and not,
in S, do A, he logically must be assenting insincerely.'

R. M. Hare, Moral Thinking (Oxford: Clarendon Press, 1981), 21.

I shall come back to this definition later on. But first, something more general about
the theories I am attacking. It is characteristic of those I have mentioned, and others
inspired by them, to suggest that the making

end p.190

of any sincere moral judgement requires the presence of individual feeling, attitude,
or intention, and thus goes beyond 'description' or 'assertion of fact'. It was
recognized, of course, that the language contains many terms like 'courage' or
'justice' designed for description as well as moral judgement, but it was said that their
'descriptive' content could not reach all the way to moral evaluation, the speaker's
feelings or commitments to action having to be added if that were to be on the scene.
Hence the apparently unquestionable distinction between 'descriptive' and
'evaluative' language, more or less taken for granted in much of contemporary ethics.
In early versions of these theories it was suggested that only a demand for
consistency set any limits on the classes of actions to which words such as 'morally
good' or 'morally bad' could be applied. So the extra feature supposedly involved in
moral judgement could stand on its own, ready to form the core of alien moral
systems confronting, or even directly contradicting, our own, and if no linguistic

device existed for expressing 'moral approval' or 'moral disapproval' in their purity,
this was held to be merely an accident of language. Thus, these early theories were
radically subjectivist, allowing the possibility even of bizarre so-called 'moral
judgements' about the wrongness of running around trees right-handed or looking at
hedgehogs in the light of the moon, and thus opening up limitless possibilities of
irresolvable moral conflict. Nowadays it is commonly admitted, I believe, that there is
some content restriction on what can intelligibly be said to be a system of morality.
Moreover, Hare himself has suggested that a fairly tight form of utilitarianism can
actually be obtained from universalized prescriptivism.

See R. M. Hare, Freedom and Reason (Oxford: Clarendon Press, 1963).

So it is not the old battle against a 'free for all'subjectivism that I want to fight. The
mistake that I referred to in my title is one I claim to find in the later, as in the earlier,
versions of non-cognitivism. Even if the very tightest limitations on 'descriptive
content' were accepted—even Bentham's suggestion that when used in conjunction
with the greatest happiness principle, words such as 'ought' and 'right' have meaning
and otherwise not—'description' would still not, according to these theories, reach all

end p.191

the way to moral judgement. Someone convinced of the utility—or whatever—of
certain kinds of action would not—indeed could not—straightforwardly and with
sincerity make the judgement about their moral goodness unless he found in himself
the right feelings and attitudes, or was ready to take the step of committing himself to
act in a particular way. For moral evaluation, something 'conative'had to be present
as well as belief in matters of fact.
What all these theories try to do, then, is to give the conditions of use of sentences
such as 'It is morally objectionable to break promises', in terms of something that
must be true about the speaker. He must have certain feelings or attitudes; he must
commit himself to acting in a certain way; he must at least feel remorse if he does not
so act. Meaning was thus to be explained in terms of a speaker's attitude, intentions,
or state of mind

. And this opened up a gap between moral judgements and

assertions, with the idea that truth conditions give, and may exhaust, the meaning of
the latter but not the former. Thus it seemed that fact, complementary to assertion,
had been distinguished from value, complementary to the expression of feeling,
attitude, or commitment to action. Propositions about matters of fact were assertible if
their truth conditions were fulfilled, but moral judgements, through conditions of
utterance, were essentially linked to an individual speaker's subjective state.
It is this kind of thing that seems to me all wrong. That is what I intended in
suggesting that moral subjectivism 'rests on a mistake'. So what, then, is the
mistake? It is the mistake of so construing what is 'special' about moral judgement
that the grounds of a moral judgement do not reach all the way to it. Whatever
'grounds' may have been given, someone may be unready, indeed unable, to make
the moral judgement, because he has not got the attitude or feeling, is not in the
'conative'state of mind, is not ready to take the decision to act: whatever it is that the
theory says is required. It is this gap between ground and moral judgement that I am
denying. In my view there are no such conditions on moral judgement and therefore
no such gap.
It was not, however, a fit of collective madness that seized moral philosophers in the
thirties, and still grips them today. Their theories were devised to take account of
something that really is a feature of

end p.192

moral judgement: the 'action-guiding' character of morality, which Hume had insisted
on and taken as the foundation of his moral philosophy. Morality, Hume had said, is

necessarily practical, serving to produce and prevent action, and I shall call this
'Hume's practicality requirement'.

See David Hume, A Treatise of Human Nature (1739-40), iii. ii.

Nor am I denying that his demand must be met. My contention is rather that the
theories I am attacking tried to meet it in the wrong way. This, substantially, is what
this paper is about.
If I am to prove my thesis I must, of course, produce an alternative to the non-
cognitivist way of showing that moral judgement is essentially 'action-guiding'. So
what is my own account of the matter? It is, to state it briefly, that Hume's demand is
met by the (most un-Humean) thought that acting morally is part of practical
rationality.
Now I am quite aware that to make this suggestion will seem most foolhardy: a case
of putting one's head, philosophically speaking, into the lion's mouth. For is it not
difficult to establish even coincidence between moral and rational action? What, after
all, about those problem cases where justice or charity forbids the only way out of a
tight corner, and the life of the agent may even be at stake? Isn't the demonstration
of the rationality of just action a problem with which David Gauthier, for instance, has
been wrestling for years, with great energy and skill?

See D. Gauthier, Morals by Agreement (Oxford: Clarendon Press, 1986).

And isn't this the fence at which I myself have repeatedly fallen, trying now this way
now that of getting over—from 'Moral Beliefs'in

1958

to 'Morality as a System of

Hypothetical Imperatives' in

P. R. Foot, 'Moral Beliefs', Proceedings of the Aristotelian Society, 59 (

1958-9

), and 'Morality as a

System of Hypothetical Imperatives', Philosophical Review, 89/3 (July

All of this is true, and if I am hopeful of greater success this time round it is because I
think I now see why I couldn't have managed it before. Roughly speaking it was
because I still held a more or less Humean theory of reasons for action, taking it for
granted that reasons had to be based on an agent's desires. To be sure, in another
article, 'Reasons for Actions and Desires',

Proceedings of the Aristotelian Society, suppl. vol. 46 (