COMMENSAL RESPONSES

The view that 'ought' cannot be deduced from 'is', credited to Hume as a major insight into the nature of morality, is surprisingly easy to refute.

Here we have a case of deducing 'ought' from 'is'. The conclusion follows, because 'ought not' is analytic to 'evil'.

'Ah, but that's just what is wrong with the example: the premise is not a pure "is"; it contains an "ought", though this does not appear explicitly.' This is true, of course; the inference would not be valid otherwise. Still, the example shows that the is/ought principle will not do as it stands.

Why does the principle seem as plausible as it does? It relies on a familiar principle of logic that the conclusion of an argument must not contain material that is not in the premises (This is not a precise, formulation, but I shall not attempt a better one.) Hume's reason for questioning the inference from is to ought is that 'this ought... expresses some new relation or affirmation' and 'it seems altogether inconceivable how this new relation can be a deduction from others [expressed by is] which are entirely different from it' (Treatise, ed. Selby-Bigge, 469). However, the writers criticized by Hume did not deduce their 'ought' conclusions merely from the word 'is'. It would be absurd to suppose that anything can follow from this word just by itself: only when 'is' is part of a predicate (such as 'is square' or 'is evil') does it make sense to speak of implication and inference. But, as we have seen, it is easy to give examples of 'is'-predicates from which 'ought' conclusions follow.

Is there a better way of expressing what Hume meant, or what is usually meant when the principle is cited ? It might be thought that the terminology of 'facts' versus 'values', or 'description' versus 'prescription' is more suitable for this purpose. The essential principle, on this view, is that we cannot deduce a value statement from a factual one, or a prescriptive from a descriptive one. And although (1) is, admittedly, an 'is' statement, it is not - according to this view - a factual or descriptive one.

An obvious difficulty of this view is that, according to the normal use of these words, 'description' and 'fact' are applicable to value statements no less than to others. 'He described what they are doing as "evil"' is a perfectly normal remark; and similarly with the word 'fact'. Perhaps what is meant can be better expressed in terms of objective facts. Hume himself allows that there is 'a matter of fact' when we describe 'any action or character to be vicious'; but, he says, the fact in question 'lies in yourself, not in the object' (469). To prove that it does not lie in the object, he challenges the reader to 'examine [an action] in all lights' to 'see if you can find that matter of fact ... which you call vice'. The vice, he concludes, cannot be found 'till you turn your reflection into your own breast...'. It is not, in this sense, an objective fact.

However, the action chosen by Hume as an example of - 'wilful murder' - draws attention to another difficulty about the distinction in question. There is a third alternative to those he puts before us: to 'find the vice' in his example we need neither to examine the action nor to look into our breasts, but only to be aware of the meaning of 'murder'. It is part of the meaning of this word - the part that distinguishes it from mere 'killing' - that the action is morally wrong. Hence the inference from 'murder' to 'wrong' (from 'is a murder' to 'ought not', etc.) is of the same kind as that from (1 ) to (2).

It might now be admitted that the essential is/ought (fact/value, etc.) principle is obscured by the use of such words as 'murder' in the premise. What is really meant, it might be said, is that an evaluative conclusion cannot follow from a value-free description. Such words as 'murder' are value-laden; but if we are given a description in value-free terms, then (on this view) the break between fact and value will be apparent.

Let us see how this would work with the example of murder. A kills B in order to take his money, and he wants the money to enable him to improve his standard of living. This seems as close as we can get to a value-free, 'purely factual' description, containing no terms that, by themselves, entail a moral ought or ought not. 'A kills B' is neutral in this sense, until we know the killer's motive; and 'A wants B's money' and 'A wants to improve his standard of living' are likewise neutral. Yet it follows from the description as a whole that the action was one of murder. The description is, indeed, that of a paradigm case of murder: a perfect example of what is meant by this word. We now have a two-stage refutation of the fact/value principle: first going from value-free premises to a value-laden description and then from the latter to an 'ought' conclusion.

'But the very fact that we could arrive at an ought conclusion proves that our premises could not really have been value-free.' In a sense this is true, of course. Taking the descriptions of A's action and the motives behind it all together, we no longer have a value-free description. The complete description is, as I said, that of a paradigm case of murder, and 'murder' is not a value-free word. Nevertheless the individual descriptions that make up the complete one are, as we saw, value-free; and so we are left with an entailment from value-free to value-laden.

But, it may still be asked, what kind of inference is involved here? In proceeding from 'evil' to 'ought not', and from 'murder' to 'wrong', we were relying on the traditional method of analytic reasoning: pointing out that 'ought not' is analytic to 'evil', etc. But this is not so when we proceed from the description of A's action and motive to the description 'murder'. There is, in this case, no particular word that has 'murder' as part of its meaning.

However, though the connection is not analytic, it still one of meaning. The connection is of a kind that is constantly in play in the use of language. From the fact that a person is male, over 50, and has never been married, it follows that the description 'bachelor' is applicable; from the fact that something is the same colour as this tomato, it follows that it can be described as 'red'; from the fact that two people are throwing dice and the first one to reach a hundred is declared the winner, and they are doing this for amusement, it follows that they are playing a game; and from the fact that A killed B in the way described it follows that A committed a murder. What we are talking about are sufficient conditions, on the basis of which such words are applied. Here is a familiar kind of logic, which is not that of analytic reasoning, but without which language would be impossible.

Taking my cue from Hume, I have used 'murder' as an example; and if my argument has been correct, this example is sufficient to refute the is/ought principle. Let us, however, consider how the inference from 'is' to 'ought' would work in the case of other offences, such as stealing, lying and promise-breaking. It is well known that the wrongness of such actions - the inference from 'stealing', say, to 'ought not' - is subject to a ceteris paribus qualification. In his article 'How to Derive "Ought" from "Is"', John Searle dealt with this by adding to his premises the statement 'Other things are equal'. (Phil. Review 1964, reprinted in W D Hudson, ed., The Is / Ought Question (Macmillan 1969). Searle took a different course in a later discussion. See Speech Acts, ch. 8.) But some critics objected that with this statement the premises were no longer free of evaluation, since a value judgement - an ought, in effect - was needed to decide that other things are, indeed, equal. (See papers by J. E. McLellan and B. P. Komisar and by James Thomson in Hudson, op. cit.)

Here one may be misled by the phrase 'other things being equal'. This may seem to refer to other moral considerations that need to be evaluated before one can proceed to an 'ought' conclusion; and this would support the objection. Thus there might be moral considerations against keeping one's promise and, perhaps, still other moral considerations (besides the promise itself) in favour of keeping it. And in such cases the competing considerations will, indeed, need to be evaluated in order to decide what is right. But this is not what is usually meant by 'other things being equal'. What is usually meant is the absence of other things - of morally relevant features that might interfere with the one existing obligation. In typical cases it is obvious that this is so: a promise, say, has been given and there are no moral reasons against keeping it. Here the ought conclusion is seen to follow from a single relevant fact without further ado.

A failure to notice this possibility may be behind Bernard Williams's qualified endorsement of the is/ought principle. The principle is true, he says, 'if the ought is taken to be the same as the should that occurs in the practical question "What should I do?" and in the "all things considered answer to it"' (Ethics and the Limits Of Philosophy 126). This question, he argues, 'always requires us to determine what, on this particular occasion, in the light of everything, we judge the most important'. Now it is true that on many occasions the facts leave undetermined what one should or ought to do, and on those occasions evaluative judgement maybe needed if we are to come to a decision. But in other cases this is not so: there is only one relevant consideration and no occasion, therefore, to 'judge what is most important... in the light of everything'.

A similar mistake seems to be involved in Simon Blackburn's critique of the idea of 'thick' concepts. He quotes an extensive list of examples from Hume, including frugality, industry, discernment and others, ' whose names', according to Hume, 'force an avowal of their merit' ('Morality and Thick Concepts' PASS 1992, 285-6; Hume's Enquiry, ed. Selby-Bigge, 242). Blackburn objects on the ground that such descriptions can be used with a negative connotation. A person's industry, he points out, might be 'entirely misdirected' and 'a general cause of grief', and yet be industry for all that (286). Similarly, turning to Aristotelian virtues such as courage, he writes that 'we can talk of "mad courage" or "Dutch courage" without linguistic impropriety'. But what do these examples show? That industry and courage (honesty, kindness etc.) are not virtues? No: what they show, or rather illustrate, is that the 'avowal of their merit' is subject to a ceteris paribus qualification: to describe a person as industrious or an action as courageous is normally to commend him or it, but this verdict may be overridden if there are special reasons to the contrary.

Blackburn acknowledges that such descriptions are normally taken to imply 'a favourable attitude' (287), but he thinks this depends on 'a theory of what a particular speaker is doing'. Similarly, 'we might expect someone who talks of a house as containing south facing windows to be implying or inviting a favourable attitude ... yet [this phrase] is not usually thought of as a thick term'. But the last clause, so far from clinching the argument, just confirms that these cases are not alike: whereas we do not think of 'south facing windows' as a thick term, we do think of discretion and the rest as thick terms. What this means (avoiding the term of art, 'thick') is that whereas south facing windows are a matter of personal preference, this is not so in the case of the virtues. No one would suppose that the description of courage as a virtue is a matter of personal preference, and the same is true of the description of murder as 'vicious'. The fact that an implication is (to use Grice's term) 'cancellable' for special reasons in particular cases does not entail that it is not part of the meaning of the word in question.

Another way of denying the objectivity of moral facts has been to regard them as dependent on human institutions. 'Murder', for example, might be regarded as a legal term, so that the entailment from the 'pure description to the conclusion 'murder' would be the law, the action in itself being neither valid only in the context of right nor wrong.

Now it is true that 'murder', 'theft' etc. are legal terms, with artificially introduced definitions: but they are not merely that. And in general, the existence of (moral and other) value-laden concepts - murder, robbery, lying, cheating, promise-breaking and the rest - is logically prior to the introduction of legal definitions. The logical priority of moral concepts is shown by the fact that the law can be, and often is, criticised on moral grounds.

It has been thought, however, that moral concepts are themselves institution-dependent. This view has been taken especially with regards to promising. It is unfortunate that promising has figured as it has in the is/ought debate, for while the institutional view may seem plausible here, it is far less so when we come to such as murder. (Note : Mackie, however, among others, has suggested that moral concepts in general are institutional, so that recognition of the relevant moral 'facts' depends on one's endorsement of the corresponding institutions (Mackie 81)). And yet, of course, the is/ought principle must be applicable to murder no less than in the case of promising.

It is also unfortunate that the institutional view was promoted by anr opponent of the is/ought principle, for this tended to put the whole debate on a false course. In his well-known article, cited above, Searle represented the obligation of promise-keeping, together with others, as 'institutional facts' depending for their existence on 'constitutive rules' comparable to those that are constitutive of baseball and other games (in Hudson, 132). This invited the response that such derivations of 'ought' from 'is' are valid only for those who choose to 'endorse' or 'assent to' the relevant game or institution; and this endorsement would amount to an 'ought' premise, enabling an 'ought' conclusion to be drawn without contravening the is/ought principle. (Hare's response was appropriately entitled 'The Promising Game'.)

But what is meant by 'endorse' and 'assent to' when philosophers speak in this way about promising? It is sometimes held that one gives one's endorsement of such 'institutions' by participating in them, so that one would be endorsing the institution of promising by actually making promises.

Such an account would be suitable for the case of games, where participation is a matter of choice. It is up to me whether to participate in a game, and it is also open to me to withdraw from a game in which I have participated. (I might do this simply by saying 'I don't want to play any more'.) But is the same true of promising. According to Mackie, one can choose to endorse this institution by making a promise, but one can also choose to withdraw from it 'when the time comes for payment'. In this way, he argues, the obligation is institution-dependent at the time of making the promise and again at the time of keeping it (J. L. Mackie, Ethics, 70).

But how are we to imagine this? Can someone who promised say 'I don't want to play any more' when the time comes for paying up? It is a mistake to regard promises and other moral facts as institutional. There is nothing institutional about the fact that the person in my example committed a murder, nor about the fact that one is obliged to do what one promised. Someone who understands the word 'murder' must recognize that the case I described would one of murder; he cannot refuse to do so as he might refuse to participate in a game or institution. He might perhaps try to abstain from using the word 'murder', though it is hard to see how he could do this if the subject came up; in any case, his recognition of the case described as one of murder would not depend on his actually uttering the word. And similarly, someone who understands the relevant language must recognize that, in suitable situations, a person who says 'I promise to do A', or merely 'I will do A, undertakes an obligation and ought to do what he said. There is no question of endorsing or agreeing to participate in an institution; and neither could one abstain from doing so by avoiding the words 'I promise' ?

In his article Searle seems to have been bewitched by the 'performative' nature of this expression. (Austin's work on performatives had appeared not long before.) One may be tempted to the view that the performative magic, so to speak, of the expression 'I promise' is made possible by the existence of a suitable institution; and it is true of some performative statements that they work only in the context of an institution. ('I christen this child ...' is an example.) This is not so, however, with promising; and neither is the use of 'I promise' essential to, or even typical of, the giving of promises. If I say, in suitable circumstances, that I will meet you at the library tomorrow, then I have given a promise, whether or not this word was used. It would be absurd for me, having failed to turn up, to rebut your complaint ('But you promised') by denying that I this word. (Note : For further discussion see O. Hanfling, 'Promises, Games and Institutions', Procs. of Arist. Soc. 1974-75). Nor does the strength of my obligation depend, particularly, on whether I have used this word. You can strengthen my obligation by explaining to me how important it is for you that I do what I said and the word 'promise' need never be used in this conversation. We can imagine a tribe in which there is no counterpart to our 'promise', but this would not mean that promises, of varying degrees of commitment, are not exchanged there; nor would it be wrong for a visiting anthropologist to translate the relevant conversations in terms of our 'promise'.

My deduction of 'ought' from 'is' in the case of a murder is similar to one that Hare, a stout defender of the is / ought principle, anticipated and rejected. Using 'courage' as an example, Hare has an opponent of the principle speak as follows:

Hare dealt with this counter-example by claiming that it is open to anyone to abstain from the evaluation in question by using 'courageous' in a 'purely descriptive sense', to mean no more than 'disregarding one's own safety in order to preserve that of others', while making it clear that one was 'implying thereby no commendation whatever' (189).

It is not obvious how this reply is supposed to defeat the objection. Suppose that some individual or individuals did abstain from using 'courage' and other moral words with their normal meanings. Would this rescue the is / ought principle from counter-examples such as 'courage' and 'murder'? No: it would still be the case that, say, 'murder' entails 'ought not'. If someone chose to use this word (supposing that were possible) in a morally neutral sense, then it would no longer be the same word, but merely a word having the same sound: it would not mean what is meant by 'murder' and it would still be the case that this word entails 'ought not'.

Similarly, if there are languages from which the concepts of murder or (say) stealing are absent, this does not affect the entailments from 'murder' and 'stealing', as we understand these words to 'wrong' and 'ought not', any more than the truth of Pythagoras' theorem, as expressed in our vocabulary, is affected by the fact that in many languages there is no such vocabulary. Nor would it make any difference if there were a language containing a word which covered such actions as that of A killing B as described above, but without any moral implication. For such a word would not be translatable as 'murder' and its existence would make no difference so the connection of this word with 'ought not'.

Apparently Hare thought of the meanings of words as a matter of decision, akin to the way in which one's endorsement of an institution would be a matter of decision. On this view, as we saw, a person who gives such an endorsement is held to be contributing an 'ought' premise; and it might be thought that, similarly, someone who chooses to use 'murder' with its moral connotation is contributing an 'ought' premise which is not to be found in the 'mere facts' of the case.

It is not true, however, that the meanings of words, such as those at issue, are a matter of choice; nor is it open to a person to use them according to abnormal criteria, any more than to choose one's own meanings for the words 'glory' or 'a nice knock-down argument'. It was also misleading of Hare to speak of the possession of the concept of courage in hypothetical terms ('if one has the word in one's vocabulary') - as if a normal adult speaker of English could fail to have it. It is true that one might try to avoid uttering the word, and perhaps succeed in this, but this does not mean that one could fail to recognize its correct use by others. This word, like others in the moral vocabulary, has an established place in the language and no one is in the position of contributing an element of value by agreeing to use it in the normal way.

Another argument used by Hare against his imagined objector involves the word 'nigger', a word which, like 'courage', includes evaluative and non-evaluative elements in its meaning. Hare suggests that if one accepts the argument from description to evaluation in the case of 'courage', then one must be prepared to accept a parallel argument in the case of 'nigger', as follows:

But, as Hare points out, this is not an argument that his reader would be likely endorse: he 'will reply that, simply because the user of the word "nigger" is led along this path, he prefers not to use the word'.

Here again, however, Hare's hypothetical language ('if we have...') is misleading, for it is clear that 'we' - Hare himself and the readers he has in mind - do not 'have the word "nigger" in use'. It is true that if we had the word in use, then we would be led ineluctably, etc.; but the fact is that we do not, and are not. Hence the case is not like those of 'courage' and 'murder'. (To deny that we have the word 'in use' is not, of course, to say that we do not understand it: it is to say that we do not, and would not, use it to describe people. We may, however, use it, in inverted commas, to describe the language of those who do or did apply it to people.)

It might be thought, again, that whether one has the word 'nigger' in use is a matter of choice ('he prefers not to use the word'). But this is not so. If I cannot recognize an entailment from 'dark skin' etc. to 'contemptible', then I cannot (sincerely) use language that commits me to such an entailment. Suppose I found myself in a society in which the word 'nigger' is in use. Could I not, in that case, choose to use the word as others do and thereby choose to endorse the relevant evaluation ? Yes, but this endorsement would be a pretence. Asked whether I really believe that 'dark skin' etc. entails 'contemptible', I could not sincerely answer 'yes' if this is not what I believe. And what I believe is not a matter of choice.

To a large extent the use of 'nigger' has, indeed, lapsed. If the same happened to 'murder', 'courage' and the rest, then we could no longer use these words to argue from 'is' to 'ought'; but these words are alive and well and the entailments from 'is' to 'ought' remain in force. It is not surprising that these words are better survivors than 'nigger'. The latter was specific to a particular language and a particular time and place, and it was always likely that someone would question the logical connection between the elements - 'dark skin, etc.' and 'contemptible' - that make up its meaning. It was also likely that when faced with this question, a defender of the word would have appealed to deeper differences which would command more general recognition, claiming that these were behind the superficial differences. Alternatively, he might have denied that his use of 'nigger' had derogatory connotations, But such moves are not necessary in the cases of 'murder' and 'courage'.

Let us return to Hume's remarks about 'ought' and 'ought not' with which we began. There is an important difference between Hume's position, as expressed there, and that of his modern followers. Both deny that moral values are to be found among objective facts, but whereas Hume locates them in our breasts, his followers claim that they are created by us. Given the connections between morality and moral language, they are then led to the view that the relevant language is a creature of human choice, so that, for example, it is up to us, individually or collectively, whether to include such concepts as murder, courage and promising in our moral vocabulary, thereby recognizing them as moral values. Hume, by contrast, does not regard moral values as objects of choice, but as part of 'the given'.

The modern philosophers are guilty of what I shall call the Fallacy of Voluntarism; Hume is not. In this respect Hume is nearer to the truth. The truth is that our moral values, and the language in which they are expressed, are given and not invented. They are not, however, given by way of sensations in our breasts, but in the way indicated by Wittgenstein when he wrote that 'what has to be accepted, the given, is - so one could say - forms of life' (PI p. 226).

This is not to say that we are wholly passive in our moral thinking. Of course there is much room for active reasoning and persuading in the moral sphere, as in others. And obviously we are in the realm of action when it comes to acting in morally relevant ways. But the concepts and categories in which these activities are conducted, and by means of which the relevant actions are described, are not themselves creatures of human choice. Again, there is obviously such a thing as thinking critically about some of the values accepted in one's society, as in the case of sexual morality; but these discussions are conducted in terms of more fundamental values, which are not themselves candidates for acceptance or rejection.

The Voluntaristic Fallacy is not confined to ethics: it can also be found in discussions of logic. J. O. Urmson, noticing that the values 'valid' and 'invalid' are applicable to logic, but unwilling to recognize values as given, claimed that these values too depend on the preferences of particular groups of people (in A. Flew, Essays in Conceptual Analysis, 128). The question 'Are these arguments really valid?' was, he thought, 'deplorable' in the same way as the question 'Are these things really good?' in both cases, he maintained, we are dealing with human choice as opposed to objective reality. Similarly, Antony Flew, refusing to apply the 'paradigm case argument' to 'any matter of value', including inductive logic, claimed that 'it is necessary for each of us... to make our personal value commitments here' (in C. Lyas, Philosophy and Linguistics, 74). (Note : For a more recent comparison of moral with logical values, see G. F. Schueler in Mind, October 1995.) But there is no conceivable situation in which it would be up to us, collectively or individually, to make our own 'value commitments' with regard to valid reasoning.

The phrase 'personal commitment' is likely to strike a chord with many people, in the case of moral values at least. But this phrase must be understood in an appropriate way. It is true that a person may commit himself to supporting some good cause, and in that sense make a personal commitment; or he may resolve or promise to behave in accordance with some recognized value, and in that sense commit himself. But the existence of the relevant values is not a matter of personal commitment, any more than the existence of 'valid' and 'invalid' in the case of arguments. (Note : I am grateful to Peter Hacker and, especially, to David Cockburn for help with previous drafts.)

Return to Commensal Article (In Commensal 90)