From Philippa R. Foot, Virtues and Vices (Berkeley: University of California Press, 1978), pp. 99–105.
“Both these accounts of moral argument are governed by the thought that there is no logical connexion between statements of fact and statements of value, so that each man makes his own decision as to the facts about an action which are relevant to its evaluation. To oppose this view we should need to show that, on the contrary, it is laid down that some things do, and some things do not, count in favour of a moral conclusion, and that a man can no more decide for himself what is evidence for rightness and wrongness than he can decide what is evidence for monetary inflation or a tumour on the brain. If such objective relations between facts and values existed, they could be of two kinds: descriptive, or factual, premises might entail evaluative conclusions, or they might count as evidence for them. It is the second possibility which chiefly concerns me, but I shall nevertheless consider the arguments which are supposed to show that the stronger relationship cannot exist. For I want to show that the arguments usually brought forward do not even prove this. I want to say that it has not even been proved that moral conclusions cannot be entailed by factual or descriptive premises.
It is often thought that Hume showed the impossibility of deducing ‘ought’ from ‘is’, but the form in which this view is now defended is, of course, that in which it was rediscovered by G. E. Moore at the beginning of the present century, and developed by such other critics of ‘naturalistic’ ethics as Stevenson, Ayer, and Hare. We need therefore to look into the case against naturalism to see exactly what was proved.
Moore tried to show that goodness was a non-natural property, and thus not to be defined in terms of natural properties; the problem was to explain the concept of a ‘natural property’, and to prove that no ethical definition in terms of natural properties could be correct. As Frankena3 and Prior4 pointed out, the argument against naturalism was always in danger of degenerating into a truism. A natural property tended to become one not identical with goodness, and the naturalistic fallacy that of identifying goodness with ‘some other thing’.
What was needed to give the attack on naturalism new life was the identification of some deficiency common to the whole range of definitions rejected by Moore, a reason why they all failed. This was provided by the theory that value terms in general, and moral terms in particular, were used for a special function—variously identified as expressing feelings, expressing and inducing attitudes, or commending. Now it was said that words with emotive or commendatory force, such as ‘good’, were not to be defined by the use of words whose meaning was merely ‘descriptive’. This discovery tended to appear greater than it was, because it looked as if the two categories of fact and value had been identified separately and found never to coincide, whereas actually the factual or descriptive was defined by exclusion from the realm of value. In the ordinary sense of ‘descriptive’ the word ‘good’ is a descriptive word and in the ordinary sense of ‘fact’ we say that it is a fact about so and so that he is a good man, so that the words must be used in a special sense in moral philosophy. But a special philosopher’s sense of these words has never, so far as I know, been explained except by contrasting value and fact. A word or sentence seems to be called ‘descriptive’ on account of the fact that it is not emotive, does not commend, does not entail an imperative, and so on according to the theory involved. This might seem to reduce the case against naturalism once more to an uninteresting tautology, but it does not do so. For if the non-naturalist has discovered a special feature found in all value judgements, he can no longer be accused of saying merely that nothing is a definition of ‘good’ unless it is a definition of ‘good’ and not ‘some other thing’. His part is now to insist that any definition which fails to allow for the special feature of value judgements must be rejected, and to label as ‘naturalistic’ all the definitions which fail to pass this test.
I shall suppose, for the sake of argument, that the non-naturalist really has identified some characteristic (let us call it f) essential to evaluative words; that he is right in saying that evaluations involve emotions, attitudes, the acceptance of imperatives, or something of the kind. He is therefore justified in insisting that no word or statement which does not have the property f can be taken as equivalent to any evaluation, and that no account of the use of an evaluative term can leave out f and yet be complete. What, if anything, follows about the relation between premises and conclusion in an argument designed to support an evaluation?
It is often said that what follows is that evaluative conclusion cannot be deduced from descriptive premises, but how is this to be shown? Of course, if a descriptive premise is redefined, as one which does not entail an evaluative conclusion, the non-naturalist will once more have bought security at the price of becoming a bore. He can once more improve his position by pointing to the characteristic f belonging to all evaluations, and asserting that no set of premises which do not entail an f proposition can entail an evaluation. If he takes this course he will be more like the man who says that a proposition which entails a proposition about a dog must be one which entails a proposition about an animal; he is telling us what to look out for in checking the entailment. What he is not so far telling us is that we can test for the entailment by looking to see whether the premise itself has the characteristic f. For all that has yet been shown it might be possible for a premise which is not f to entail a conclusion which is f, and it is obviously this proposition which the non-naturalist wants to deny.
Now it may seem obvious that a non-evaluative premise could not entail an evaluative conclusion, but it remains unclear how it is supposed to be proved.
In one form, the theory that an evaluative conclusion of a deductive argument needs evaluative premises is clearly unwarrantable; I mention it only to get it out of the way. We cannot possibly say that at least one of the premises must be evaluative if the conclusion is to be so; for there is nothing to tell us that whatever can truly be said of the conclusion of a deductive argument can truly be said of any of the premises. It is not necessary that the evaluative element should ‘come in whole’, so to speak. If f has to belong to the premises it can only be necessary that it should belong to the premises together, and it may be no easy matter to see whether a set of propositions has the property f.
How in any case is it to be proved that if the conclusion is to have the characteristic f the premises taken together must also have it? Can it be said that unless this is so it will always be possible to assert the premises and yet deny the conclusion? I shall try to show that this at least is false, and in order to do so I shall consider the case of arguments designed to show that a certain piece of behaviour is or is not rude.
I think it will be agreed that in the wide sense in which philosophers speak of evaluation, ‘rude’ is an evaluative word. At any rate it has the kind of characteristics upon which non-naturalists fasten: it expresses disapproval, is meant to be used when action is to be discouraged, implies that other things being equal the behaviour to which it is applied will be avoided by the speaker, and so on. For the purpose of this argument I shall ignore the cases in which it is admitted that there are reasons why something should be done in spite of, or even because of, the fact that it is rude. Clearly there are occasions when a little rudeness is in place, but this does not alter the fact that ‘rude’ is a condemnatory word.
It is obvious that there is something else to be said about the word ‘rude’ besides the fact that it expresses, fairly mild, condemnation: it can only be used where certain descriptions apply. The right account of the situation in which it is correct to say that a piece of behaviour is rude, is, I think, that this kind of behaviour causes offence by indicating lack of respect. Sometimes it is merely conventional that such behaviour does indicate lack of respect (e.g. when a man keeps his hat on in someone else’s house); sometimes the behaviour is naturally disrespectful, as when one man pushes another out of the way. (It should be mentioned that rudeness and the absence of rudeness do not exhaust the subject of etiquette; some things are not rude, and yet are ‘not done’. It is rude to wear flannels at a formal dinner party, but merely not done to wear a dinner jacket for tennis.)
Given that this reference to offence is to be included in any account of the concept of rudeness, we may ask what the relation is between the assertion that these conditions of offence are fulfilled—let us call it O—and the statement that a piece of behaviour is rude—let us call it R. Can someone who accepts the proposition O (that this kind of offence is caused) deny the proposition R (that the behaviour is rude)? I should have thought that this was just what he could not do, for if he says that it is not rude, we shall stare, and ask him what sort of behaviour would be rude; and what is he to say? Suppose that he were to answer ‘a man is rude when he behaves conventionally’, or ‘a man is rude when he walks slowly up to a front door’, and this not because he believes that such behaviour causes offence, but with the intention of leaving behind entirely the usual criteria of rudeness. It is evident that with the usual criteria of rudeness he leaves behind the concept itself; he may say the words ‘I think this rude’, but it will not on that account be right to describe him as ‘thinking it rude’. If I say ‘I am sitting on a pile of hay’ and bring as evidence the fact that the object I am sitting on has four wooden legs and a hard wooden back, I shall hardly be described as thinking, even mistakenly, that I am sitting on a pile of hay; all I am doing is to use the words ‘pile of hay’.
It might be thought that the two cases were not parallel, for while the meaning of ‘pile of hay’ is given by the characteristics which piles of hay must possess, the meaning of ‘rude’ is given by the attitude it expresses. The answer is that if ‘thinking a thing rude’ is to be described as having a particular attitude to it, then having an attitude presupposes, in this case, believing that certain conditions are fulfilled. If ‘attitudes’ were solely a matter of reactions such as wrinkling the nose, and tendencies to such things as making resolutions and scolding, then thinking something rude would not be describable solely in terms of attitudes. Either thinking something rude is not to be described in terms of attitudes, or attitudes are not to be described in terms of such things. Even if we could suppose that a particular individual could react towards conventional behaviour, or to walking slowly up to an English front door, exactly as most people react to behaviour which gives offence, this would not mean that he was to be described as thinking these things rude. And in any case the supposition is nonsense. Although he could behave in some ways as if he thought them rude, e.g. by scolding conventional or slow-walking children, but not turning daughters with these proclivities out of doors, his behaviour could not be just as if he thought them rude. For as the social reaction to conventional behaviour is not the same as the social reaction to offensive behaviour, he could not act in just the same way. He could not for instance apologise for what he could call his ‘rudeness’, for he would have to admit that it had caused no offence.
I conclude that whether a man is speaking of behaviour as rude or not rude, he must use the same criteria as anyone else, and that since the criteria are satisfied if O is true, it is impossible for him to assert O while denying R. It follows that if it is a sufficient condition of P’s entailing Q that the assertion of P is inconsistent with the denial of Q, we have here an example of a non-evaluative premise from which an evaluative conclusion can be deduced. It is of course possible to admit O while refusing to assert R, and this will not be like the refusal to say about prunes what one has already admitted about dried plums. Calling an action ‘rude’ is using a concept which a man might want to reject, rejecting the whole practice of praising and blaming embodied in terms such as ‘polite’ and ‘rude’. Such a man would refuse to discuss points of etiquette, and arguments with him about what is rude would not so much break down as never begin. But once he did accept the question ‘Is this rude?’ he would have to abide by the rules of this kind of argument; he could not bring forward any evidence he liked, and he could not deny the relevance of any piece of evidence brought forward by his opponent. Nor could he say that he was unable to move from O to R on this occasion because the belief in O had not induced in him feelings or attitudes warranting the assertion of R. If he had agreed to discuss rudeness he had committed himself to accepting O as evidence for R, and evidence is not a sort of medicine which is taken in the hope that it will work. To suggest that he could refuse to admit that certain behaviour was rude because the right psychological state had not been induced, is as odd as to suppose that one might refuse to speak of the world as round because in spite of the good evidence of roundness a feeling of confidence in the proposition had not been produced. When given good evidence it is one’s business to act on it, not to hang around waiting for the right state of mind. It follows that if a man is prepared to discuss questions of rudeness, and hence to accept as evidence the fact that behaviour causes a certain kind of offence, he cannot refuse to admit R when O has been proved.
The point of considering this example was to show that there may be the strictest rules of evidence even where an evaluative conclusion is concerned. Applying this principle to the case of moral judgements, we see that—for all that the nonnaturalist has proved to the contrary—Bentham, for instance, may be right in saying that when used in conjunction with the principle of utility ‘the words ought and right and wrong, and others of that stamp, have a meaning: when otherwise they have none’.5 Anyone who uses moral terms at all, whether to assert or deny a moral proposition, must abide by the rules for their use, including the rules about what shall count as evidence for or against the moral judgement concerned. For anything that has yet been shown to the contrary these rules could be entailment rules, forbidding the assertion of factual propositions in conjunction with the denial of moral propositions. The only recourse of the man who refused to accept the things which counted in favour of a moral proposition as giving him a reason to do certain things or to take up a particular attitude, would be to leave the moral discussion and abjure altogether the use of moral terms.”
As a bonus, here is Henry Hazlitt addressing the same issue in The Foundations of Morality, pg. 33-36:
“True, it would have no claim to scientific validity, or even any claim to be a useful field of inquiry, unless it were based in some convincing way on what was or what is. But here we have stepped into the very center of an age-old controversy. Many ethical writers have contended during the last two centuries that “no accumulation of observed sequences, no experience of what is, no predictions of what will be, can possibly prove what ought to be.”1 And others have even gone on to assert that there is no way of getting from an is to an ought.
If the latter statement were true, there would be no possibility of framing a rational theory of ethics. Unless our oughts are to be purely arbitrary, purely dogmatic, they must somehow grow out of what is.
Now the connection between what is and what ought to be is always a desire of some kind. We recognize this in our daily decisions. When we are trying to decide on a course of action, and are asking advice, we are told, for example: “If you desire to become a doctor, you must go to medical school. If you desire to get ahead, you must be diligent in your business. If you don’t want to get fat, you must watch your diet. If you want to avoid lung cancer, you must cut down on cigarettes,” etc. The generalized form of such advice may be reduced to this: If you desire to attain a certain end, you ought to use a certain means, because this is the means most likely to achieve it. The is is the desire; the ought is the means of gratifying it.
So far, so good. But how far does this get us toward a theory of ethics? For if a man does not desire an end, there seems no way of convincing him that he ought to pursue the means to that end. If a man prefers the certainty of getting fat, or the risk of a heart attack, to curbing his appetite or giving up his favorite delicacies; if he prefers the risks of lung cancer to giving up smoking, any ought based on the assumption of a contrary preference loses its force.
A story so old that it is told as an old one even by Bentham2 is that of the oculist and the sot: A countryman who had hurt his eyes by drinking went to a celebrated oculist for advice. He found him at table, with a glass of wine before him. “You must leave off drinking,” said the oculist. “How so?” says the countryman. “You don’t, and yet methinks your own eyes are none of the best.”—“That’s very true, friend,” replied the oculist: “but you are to know, I love my bottle better than my eyes.”
How, then, do we move from any basis of desire to any theory of ethics?
We find the solution when we take a longer and broader view. All our desires may be generalized as desires to substitute a more satisfactory state of affairs for a less satisfactory state. It is true that an individual, under the immediate influence of impulse or passion, of a moment of anger or rage, malice, vindictiveness, or the desire for revenge, or gluttony, or an overwhelming craving for a release of sexual tension, or for a smoke or a drink or a drug, may in the long run only reduce a more satisfactory state to a less satisfactory state, may make himself less happy rather than more happy. But this less satisfactory state was not his real conscious intention even at the moment of acting. He realizes, in retrospect, that his action was folly; he did not improve his condition, but made it worse; he did not act in accordance with his long-run interests, but against them. He is always willing to recognize, in his calmer moments, that he should choose the action that best promotes his own interests and maximizes his own happiness (or minimizes his own unhappiness) in the long run. Wise and disciplined men refuse to indulge in immediate pleasures when the indulgence seems only too likely to lead in the long run to an overbalance of misery or pain.
To repeat and to sum up: It is not true that “no amount of is can make an ought.” The ought rests, in fact, and must rest, either upon an is or upon a will be. The sequence is simple: Every man, in his cool and rational moments, seeks his own long-run happiness. This is a fact; this is an is. Mankind has found, over the centuries, that certain rules of action best tend to promote the long-run happiness of both the individual and society. These rules of action have come to be called moral rules. Therefore, assuming that one seeks one’s long-run happiness, these are the rules one ought to follow.”