The Philosophy and History of the Moral ‘Ought’: Some of Anscombe’s Objections

Abstract

According to G.E.M Anscombe’s paper ‘Modern Moral Philosophy’, modern moral philosophy has introduced a spurious concept of moral obligation, and has therefore made a mistake that the Greeks, and Aristotle in particular, avoided. Anscombe argues that the modern concepts of obligation, duty, and the moral ‘ought’ are the remnants of an earlier, but post-Aristotelian conception of ethics, and that they ought to be abandoned. An examination of Anscombe’s historical and philosophical claims shows that we have no reason to take them seriously. In particular, they rest on a misinterpretation of Scholastic views on ‘ought’ and obligation.

1 The Moral ‘Ought’

G.E.M Anscombe’s paper ‘Modern Moral Philosophy’¹ has exerted considerable influence since it was published, in 1958. It is probably one of the main sources for contemporary interest in ‘virtue ethics’. Those who accept some of Anscombe’s conclusions believe that, even if she is wrong in details, she has none the less put her finger on something important about morality, or the moral ‘ought’, or moral obligation, and that she has given us good reason to doubt the legitimacy of these concepts. This view - that there is a grain of truth in her argument - is the one that I will oppose. She gives no good reason for us to abandon appeals to moral obligation. Equally, she gives us no good reason to turn to virtue ethics in the hope of doing without such appeals, since she gives us no good reason to suppose that a plausible version of virtue ethics can do without them.

Since part of Anscombe’s argument is historical, I need to examine some historical claims.² But first I will try to set out some of the puzzling aspects of her argument, so that we can see where and why the historical claims matter.

In Anscombe’s view, modern moral philosophy has introduced a spurious concept of moral obligation, and has therefore made a mistake that the Greeks, and Aristotle in particular, avoided.³ To defend her claim about the radical difference between Aristotle and modern moral philosophy, and to show that Aristotle is right, Anscombe argues that the modern concepts of obligation, duty, and the moral ought⁴ are the remnants of an earlier, but post-Aristotelian conception of ethics, and that they ought to be abandoned.⁵ Aristotle’s use of ‘ought’ and related terms does not fit our conception of moral obligation, because he uses ‘ought’ only in a non-moral and functional sense (as in ‘machinery ought to be oiled’, because it will run badly without oil). This functional and non-moral sense of ‘ought’ is to be contrasted with the moral sense, which is not to be found in Aristotle.

Since Aristotle’s use of ‘ought’⁶ lacks this moral sense, the term ‘moral’, used in this sense, has no place in a description of Aristotle’s views.

If someone professes to be expounding Aristotle and talks in a modern fashion about 'moral' such-and-such, he must be very imperceptive if he does not constantly feel like someone whose jaws have somehow got out of alignment: the teeth don't come together in a proper bite. (2)

Hence we shoud not attribute any special moral sense to the terms in which Aristotle describes the virtues.⁷

The main features that Anscombe attributes to the moral ‘ought’ are these: (1) ‘Ought’ and related terms (‘should’, ‘must’, etc.) now have a special so-called moral sense. (2) This special sense involves some absolute verdict. (3) The terms have acquired this special sense because they were previously equated with ‘is obliged’ in the legal sense.

The sense she attaches to ‘absolute verdict’ is clarified by her contrast with the non-moral ‘ought’ that we use when we say that a machine ought to be oiled regularly.

The terms “should“ or “ought" or “needs" relate to good and bad: e.g. machinery needs oil, or should or ought to be oiled, in that running without oil is bad for it, or it runs badly without oil. According to this conception, of course, "should" and "ought" are not used in a special "moral" sense when one says that a man should not bilk. (5)

Anscombe does not explain how we are to understand the non-moral sense of ‘ought’ in ‘a man ought not to (should not) bilk’, or ‘one ought not to do unjust actions’. The example of oiling the machine suggests that ‘ought to be done’ is equivalent to ‘needs to be done’ or ‘must be done’, and that such a judgment is true because of some fact about the end for which the action is needed.

To explain how the moral sense differs from the non-moral sense, Anscombe mentions obligation, being bound, and being required in the sense in which one can be obliged or bound or required by law.

But they [sc. ‘ought’ etc] have now acquired a special so-called “moral” sense-i.e. a sense in which they imply some absolute verdict (like one of guilty / not guilty on a man) on what is described in the “ought” sentences used in certain types of context: … The ordinary (and quite indispensable) terms “should,” “needs,” “ought,” “must”- acquired this special sense by being equated in the relevant contexts with "is obliged," or "is bound," or "is required to," in the sense in which one can be obliged or bound by law, or something can be required by law. (5)

First she mentions an absolute verdict. Then she mentions being obliged and so on in a legal sense. One might ask what sense this is. It is not distinctive because it imports necessity; the non-moral uses also import a type of necessity, the type that is explained teleologically. The reference to the absolute verdict suggests that this is where the distinctively legal sense of ‘bound’ and ‘required’.

An absolute verdict suggests some contrast with a merely relative or conditional verdict. But it is not clear that the non-moral ought is not absolute in the same way. If I say ‘You must oil the machine now’, I do not mean the same as ‘You must oil the machine now if you want to keep it running’. You would act in accordance with the conditional if you did not care about keeping the machine running, but you would act in accordance with the non-conditional judgment only if you oiled the machine. As far as this feature of the non-moral ought goes, it is just as absolute as the moral ought.

Perhaps, then, we might supplement Anscombe. We may explain ‘absolute’ by saying that the moral ‘ought’ does not depend on any further ground such as ‘because without the oil the machine will not run well’. In the case of oiling the machine, the ground imposes some limit on the scope of the ought; if you have no reason to make the machine run well, it is false that you ought to oil it. The moral ought, however, may be taken to be unlimited in scope, because it rests on no ground that would limit its scope to some circumstances rather than others.⁸ There is no escape from the requirement, because the ground for the requirement is inescapable. The requirement to oil the machine applies only to someone who has reason to make the machine run well. But moral requirements rest on a ground that applies to everyone.

This explanation raises a question about whether the Aristotelian virtues can provide an inescapable ground. If they could, they could ground an absolute verdict. Since Anscombe assumes that they cannot ground an absolute verdict, she apparently assumes that they cannot provide an inescapable ground. I will express this assumption by saying that they provide only optional grounds that limit the scope of the ought. We are not required to have the concern for the ground that defines the scope of the non-moral ought, and hence the non-moral ought coming from the virtues does not provide a requirement that constitutes an absolute verdict.

I offer this suggestion about Anscombe’s meaning with some hesitation, since she does not explain the contrast that she has in mind beyond her remarks on the absolute verdict and on requirements and so on in the legal sense. But at least her argument presupposes an ‘Aristotelian gap’ – a gap that separates whatever sort of ‘ought’ can be grounded in the Aristotelian virtues from the moral ‘ought’. If there were no such gap, we would not need any discussion of the moral ‘ought’ apart from a discussion of the Aristotelian virtues.

To explain why the moral ‘ought’ arises from the legal use of ‘obligation’ and ‘obliged’, Anscombe suggests that the presence of a law explains the absolute character of the moral ‘ought’. Neither morality nor law leaves it up to us whether we comply with them. A law does not say ‘If you want to keep out of prison or avoid a fine, drive on the right’. If this were all it said, we would not be violating it by driving on the left and being willing to go to prison or pay a fine. Anscombe suggests that moral obligations share this feature with legal obligations because they are legal obligations. Moreover, she supposes that my being obliged by a law is identical to my being obliged by the command of a legislator who is not me. If we speak of an obligation without a legislator, we make as little sense as we would if we spoke of a verdict without a law and a judge (2).⁹

2 Historical Sources of the Moral ‘Ought’

In order to identify the post-Aristotelian conception of ethics that has shaped our concept of morality and the moral ought, and in order to explain why ‘ought’ acquired a specifically moral sense that relies on a legal concept of obligation, Anscombe appeals to the influence of Christianity with its ‘law conception of ethics’.¹⁰ Since Aristotle precedes the legal framework of Christianity, he does not use ‘ought’ in the moral sense.

This historical background explains our modern use of ‘ought’ in the moral sense, even though ‘we’ (i.e., the people Anscombe is addressing) no longer believe in a divine legislator.¹¹ Since we are post-Christian in this respect, we use the term in the moral sense, but our use rests on presuppositions that have been generally abandoned. Mediaeval Christians had a good reason to believe in genuine moral obligation, since they treated moral principles as products of divine legislation. Since modern philosophers try to keep moral obligation without divine legislation, they have no reason to believe there is any such thing.

From this brief historical story we can see a conclusion that Anscombe might draw. In her view, obligation is being bound or required by a law, and the word ‘ought’ in its moral sense has the sense of ‘obligation’. Hence, when we say ‘You ought to do x’, we mean ‘You are bound or required by a law to do x’. The sense of ought-judgments, then, is clear, but they are all false if it is possible to be bound or required as by a law only if there is a divine legislator.¹²

3 Why Hume was Right

This, however, is not Anscombe’s conclusion. If ought-judgments are all false, they are all capable of being true or false. Moreover, if the relevant law requires you to do something, you ought to do it Anscombe does not draw this conclusion, because she agrees with Hume’s view (as she interprets it) that ought-judgments cannot be true or false, and that ‘ought’ cannot be derived from ‘is’.

So Hume discovered the situation in which the notion ‘obligation’ survived, and the word ‘ought’ was invested with that peculiar force having which it is said to be used in a ‘moral’ sense, but in which the belief in divine law had long since been abandoned; for it was substantially given up among Protestants at the time of the Reformation. The situation, if I am right, was the interesting one of the survival of a concept outside the framework of thought that made it a really intelligible one. (6).

Hume does not grasp the cause, but he recognizes the effect.

If the conclusion that I mentioned were Anscombe’s conclusion, she would have no reason to agree with Hume. If ‘ought’ means ‘bound or required as by a law’, ‘ought’ can be derived from ‘is’, and moral judgments using the moral ‘ought’ are capable of being true or false, since they are all false. Hume is right, because the moral ‘ought’ has some sense that prevents us from attributing truth or falsity to these moral judgments. Anscombe says it has a ‘peculiar force’, or ‘mere mesmeric force’, or aptness ‘to have a strong psychological effect’. These remarks suggest that she accepts a non-cognitive analysis of the moral ‘ought’. If we combine these remarks with the connexion she sees between moral oughts and commands, we may reasonably conclude that the form of non-cognitivism that she prefers is prescriptivism. The prescriptive force of ‘ought’ belonged to it when speakers and hearers believed in a divine legislator, because their beliefs about what a divine legislator required had a strong psychological effect. But the force is all that remains once the belief in a divine legislator has been given up.

To see that Anscombe holds this non-cognitive account of ‘ought’, we may consider her reply to a possible objection to such an account.

This word ‘ought’, having become a word of mere mesmeric force, could not, in the character of having that force, be inferred from anything whatever. It may be objected that it could be inferred from other ‘morally ought’ sentences: but that cannot be true. The appearance that this is so is produced by the fact that we say ‘All men are phi’ and ‘Socrates is a man’ implies ‘Socrates is phi’. But here ‘phi’ is a dummy predicate. We mean that if you substitute a real predicate for ‘phi’, the implication is valid. A real predicate is required; not just a word containing no intelligible thought: a word retaining the suggestion of force, and apt to have a strong psychological effect, but which no longer signifies a real concept at all. (8)

Anscombe denies that we can infer moral oughts from premisses that contain moral oughts, on the ground that ‘ought’ is not a real predicate. From ‘All men are F, and ‘Socrates is a man’ we can infer ‘Socrates is F’, provided that for ‘F’ we substitute a real predicate. But, Anscombe asserts, we cannot argue in the same way from ‘You ought not to hate your enemy’ and ‘A is your enemy’ to ‘You ought not to hate A’, if the ‘ought’ is a moral ‘ought’; for a moral ‘ought’ does not introduce a real predicate.¹³

This assertion of Anscombe’s contains an implicit answer to the argument against non-cognitivism that Geach presents, with a reference to Frege.¹⁴ According to the ‘Frege-Geach’ argument, the obvious validity of moral arguments such as the one that I have just given counts against non-cognitivism. Since the conclusion of a valid argument is true if the premisses are true, we can attribute validity to an argument only if we take the premisses to be capable of being true or false. Hence the validity of moral arguments implies that moral judgments are capable of being true or false.

Anscombe’s answer denies the validity of moral arguments of the sort that Geach relies on. Since the moral ‘ought’ is not, as she puts it, a real predicate, it cannot be a necessary constituent of any valid moral arguments. If Anscombe takes this view of the moral ‘ought’, she seems to face similar questions about occurrences of ‘ought’ in non-assertive contexts, when I wonder whether I ought, doubt whether I ought, ask whether I ought, and so on. How, we might ask, can ‘a word of mere mesmeric force’ appear to be intelligible in all these contexts? Perhaps her view is the same as her view of ostensible moral arguments; we think we are asking genuine questions, but we are not, because our supposed question lacks a real predicate. Since Anscombe embraces these extreme conclusions about the sense of the moral ‘ought’, she believes that ‘mesmeric force’ is the whole of its meaning, when it is used by someone who does not derive it from divine commands.

In her view, the concept marked by ‘ought’ in its moral sense was intelligible for those who believed in the relevant sort of legislator, but it is no longer intelligible for those who do not believe in the relevant sort of legislator. It follows that, in her view, the concept of moral obligation that was used by mediaeval Christians is the same concept as the one we use without believing in divine law. That is why she speaks of ‘the survival of a concept outside the framework of thought that made it a really intelligible one’ (31).

4 What Survives? A Concept or a Word?

From this survey of Anscombe’s historical argument, we can attribute to her some conclusions that deserve further discussion:

• 1. For those who held the law conception of ethics, the concept introduced by the moral ‘ought’ essentially included obligation by divine law.¹⁵

• 2. For Anscombe’s presumed readers, as for Hume, the concept introduced by the moral ‘ought’ is purely prescriptive.

• 3. This concept is no longer intelligible ‘outside the framework of thought that made it a really intelligible one’. This framework was the law conception.

• 4. This concept has survived, and therefore we use the same concept that was used by believers in divine legislation.

• 5. We should give up the use of the moral ‘ought’ if we can.

These different conclusions constitute Anscombe’s case against the continued use of the moral ‘ought’.

Her different claims are inconsistent. (1), (2), and (4) are inconsistent, because (1) and (2) imply that the relevant concept has not survived, but (4) implies that it has survived. Moreover, (2) and (3) are inconsistent, because a purely prescriptive concept, as implied by (2) is perfectly intelligible. If Anscombe’s case is inconsistent, we will need to see whether a cogent argument still remains once we decide which of the inconsistent claims should be rejected.

We can see that Anscombe has not explained how the relevant concept, as she understands it, has survived, if we recall her endorsement of Hume. If a mediaeval Hume would have been right to offer a purely prescriptive analysis of the moral ‘ought’, Anscombe’s account of that concept, as used by believers in a law conception, would be wrong. If her account of that concept is correct, but Hume’s purely prescriptive analysis (as Anscombe intrerprets it) of the moral ‘ought’ is also correct, the moral ‘ought’ introduces two different concepts; for believers in a law conception a descriptive analysis is right, and for Hume’s contemporaries a prescriptive analysis is right. If the prescriptive analysis is right, law and obligation provide the criteria for the use of ‘ought’, but do not provide its meaning (just as sweetness and ripeness provide criteria for the use of ‘good’ for strawberries, but not its meaning). If the descriptive analysis is right, ‘ought’ has a descriptive meaning, and a prescriptive force that results from its use in the right contexts. Whichever analysis is right, only one can be right if the concept is the same.

Two examples that Anscombe provides to illustrate her claim about ‘ought’ help us to see why that claim is puzzling. First, she compares the survival of the allegedly moral sense of ‘ought’ with an imagined case in which (she says) the notion ‘criminal’ survives after criminal law and criminal courts have been abolished and forgotten. But how, we might ask, could it survive if criminal laws and courts have been forgotten? If they were merely abolished, we might say that ‘criminal’ could still bear the sense ‘deserving to be punished by criminal law’. But if criminal laws have been forgotten, how can the word be used with the same sense? If it does not bear the same sense, how can the notion survive?

Secondly, she compares the view of those who claim to use ‘ought ‘in its moral sense, but without divine law to a view that would speak of verdicts without laws and judges.¹⁶ In a legal verdict, a competent authority (a judge or a jury) pronounces on a matter of fact – whether someone did or did not violate the law by (e.g.) stealing. To pretend to pronounce a legal verdict on an action without any law or judge or jury would be to misunderstand what a legal verdict is. Anscombe acknowledges that in this case we would be using the word ‘verdict’ without its meaning. If the same meaning is necessary for the same concept, anyone who uses ‘verdict’ without reference to any law or judge does not use the concept of a verdict. Anscombe claims, however, that when the moral ‘ought’ is used without reference to a divine law, a concept survives ‘outside the framework of thought that made it a really intelligible one’ (6). Her conclusion about the use of ‘verdict’ – that the word survives without the concept – conflicts with her conclusion about ‘ought’.

The second inconsistency in Anscombe’s conclusions – that the current use of the moral ‘ought’ is not really intelligible – results from her acceptance of a non-cognitive analysis of the moral ‘ought’, as it is currently used. Such an analysis does not make the concept unintelligible; ‘boo!’ and ‘hurrah!’ are not unintelligible. If Anscombe had changed her mind, and decided that the descriptivist analysis of the use of ‘ought’ by adherents of the law conception is still the right analysis of our use of ‘ought’, she would still not have shown that the concept is unintelligible; it would be perfectly intelligible, and all our moral judgments using ‘ought’ would be false.

So far, then, we have no reason to believe that Anscombe has raised a serious difficulty for the moral ‘ought’. Her argument to show that something is wrong with it leads her to inconsistent conclusions.

5 Why Non-Cognitivism?

If her conclusions are inconsistent, perhaps we should reject some and retain others, and the ones that we retain may still constitute a good case against the moral ‘ought’. What, then, should we reject, and what should we retain?

We might abandon the claim that the relevant concept has survived from the time when the law conception was dominant. We can then simply concentrate on the prescriptivist analysis that Anscombe accepts for our present moral use of ‘ought’. To see why she accepts this analysis, we should ask why she believes Hume is right to deny that the moral ‘ought’ can be derived from ‘is’.

She reaches this conclusion only after considering arguments against hasty Humean conclusions. In her view, Hume’s claim raises three different problems:

• 1. The transition from ‘is’ to ‘owes’ in the case of ‘A supplied potatoes to B’ and ‘B owes money to A’ is a transition from one matter of fact to another in which the ‘is’ fact is ‘brute’ relative to ‘owes’. In normal circumstances the combination of the relevant relatively brute facts provide sufficient conditions for the truth of ‘owes’, and hence it is a fact that B owes money to A.

• 2. The transition from ‘is’ to ‘needs’ in the case of ‘this plant will die without water’ and ‘this plant needs water’ is a transition from one matter of fact to another. The ground for ‘needs’ is the fact that the plant will not flourish without water. In this case, as in the previous case, we can use ‘ought’ instead of ‘owes’ or ‘needs’, but our use of it does not affect the factual character of the conclusion.

• 3. The transition from ‘is’ to ‘ought’ in the moral sense is a transition from a fact to a non-fact.

In the case of the first two ‘oughts’ we pass from one fact to another because neither fact implies anything about the motivation of the agent (in first-person cases).¹⁷ Anscombe agrees that we will not do anything about what the plant needs or about what we need, unless we have the appropriate desire, but she argues that this makes no difference to the meaning of ‘ought’, or to the truth of the ought-judgment. In the third case Anscombe argues that since ought-judgments are purely ‘mesmeric’, and therefore not about matters of fact, we cannot derive ‘ought’ from ‘is’.¹⁸ She implies that unless we already accept a non-cognitive account of ‘morally ought’, we have no reason to accept Hume’s assertion. Hume does not provide an argument for a non-cognitive account of ‘ought’, but presupposes such an account.

Whether or not this is a correct account of Hume,¹⁹ Anscombe’s discussion shows that she does not argue for a non-cognitive account of the moral ‘ought’, but simply asserts it. We might even wonder whether she assumes such an account purely for the sake of argument, because she is writing in 1958 and non-cognitivists are all around.²⁰ This is probably not her reason, however. If she were starting from non-cognitivism, one might wonder why she so firmly rejects a non-cognitivist analysis of non-moral ‘oughts’ and of ‘good’. Her view does not seem to be that non-cognitivism should be taken as the starting point, but that the moral ‘ought’ is an important exception to the general falsity of non-cognitivism. She does not accept Hume’s argument about ‘is’ and ‘ought’ in general. She only accepts his conclusion about the moral ‘ought’.

We can state this conclusion in a different way. Anscombe’s discussion of brute facts and of non-moral oughts shows that, if we do not already accept a non-cognitive analysis of the moral ‘ought’, the non-moral uses of ‘ought’ should suggest to us a plausible account of the moral use as well. The account that Anscombe offers of the non-moral uses may be available for the moral use, unless we already assume a non-cognitive analysis of the moral use. None of the considerations mentioned by Hume, therefore, gives us any reason to accept a non-cognitive analysis of the moral use.

Anscombe suggests that one only needs to state a non-cognitive analysis of the moral ‘ought’ to see that we should give up using this ‘ought’ if we can.²¹ She declares that it has been shown to ‘lack content’. But iIf she means by this that it has been shown to lack descriptive meaning, that is no reason to abandon it, if we accept a non-cognitive analysis.

Even if the correctness of a non-cognitive analysis of the moral ‘ought’ were a good reason for abandoning its use, Anscombe would have given us no reason to abandon its use; for she gives us no reason to accept a non-cognitive analysis. She assumes that such an analysis is correct, but does not defend it.

6 The Law Conception and the Meaning of ‘Ought’

Anscombe would perhaps have a better case if she gave up her non-cognitivist account of our present use of ‘ought’. If she held that we still mean by ‘morally ought’ what believers in a law conception meant by it, and that all our judgments about what people morally ought to do are therefore false, she would have given a good reason for abandoning the use of ‘morally ought’. We might even suppose that this what she ‘really means’, or at any rate ought to have said, and that her acceptance of non-cognitivism about our present use of ‘ought’ is an aberration. If we are evaluate this revision of Anscombe’ argument, we need to look more closely at her account of the law conception and of the concept of the moral ‘ought’ that belongs to it.

Since Anscombe speaks of ‘Christianity, with its law conception of ethics’, we might suppose that anyone who accepts the Christian belief in a divine moral law thereby accepts a law conception of ethics.²² This is how Anscombe sometimes expresses her view.

To have a law conception of ethics is to hold that what is needed for conformity with the virtues failure in which is the mark of being bad qua man … that what is needed for this, is required by divine law. (6)

She is right to say that this view is part of the Christian belief in divine moral law.

This true claim, however, is insufficient for Anscombe’s argument. She wants to explain how ‘ought’ was ‘equated in the relevant contexts with “is obliged” or “is bound” or “is required to”, in the sense in which one can be obliged or bound by law, or something can be required by law’.²³ This equation of these concepts in the relevant sense does not necessarily result from a law conception, as she initially describes it. The initial description of the law conception says only:

• (1) What is needed for the virtues is required by divine law.

Such a view does not equate ‘ought’ with ‘obliged to by a law’. For the equation of the concepts we need:

• (2) The concept ‘morally ought’ (i.e. ‘needed for the virtues’) is the same concept as ‘morally obligatory’ (i.e. ‘required by divine law’).

If these two concepts are identical, someone who uses ‘morally ought’, but does not believe in a divine moral law, must regard all ‘morally ought’ judgments as false.

To show that Christian moralists (before the Reformation, at any rate) believe that these two concepts are identical, Anscombe needs to show more than that they believe that actions needed for the virtues are required by divine law. She needs to show that they understand ‘morally ought’ in the way she describes. But she does not argue for this claim. Moreover, no plausible argument can be given for it. Among mediaeval Christian moralists we can distinguish (roughly speaking) naturalists and voluntarists. Naturalists, including Aquinas and Suarez, maintain that actions are morally right (honesta) and that we ought (debere) to do them independently of whether they are required by divine legislation. Voluntarists, including Scotus and Ockham, deny that actions are morally right, or that we morally ought to do them, independently of divine legislation. Neither naturalists nor voluntarists take the concept ‘morally ought’ to be identical to the concept ‘required by divine legislation’. Voluntarists sometimes argue that naturalists are wrong to suppose that what we morally ought to do is independent of divine legislation; but they do not argue that naturalists are wrong to distinguish these two concepts.

To illustrate the distinction between the two concepts, we may observe how Aquinas describes the the moral ought (debere) in contrast to any legal obligation:

The legal due (debitum) is that which one is constrained by law (lege adstringitur) to render; and this due is chiefly the concern of justice, which is the principal virtue. On the other hand, the moral due is what someone owes (debet) out of the rightness (honestas) of virtue: and since a due implies necessity, this kind of due has two degrees. For one due is so necessary that without it rightness in morality (honestas morum) cannot be maintained; and this has more of the character of due. ... There is another due that is necessary in the sense that it conduces to greater rightness, although without it rightness may be maintained. This due is the concern of generosity, affability or friendship, or the like, all of which … have little of the character of the due. (Aquinas, Summa Theologiae 2–2 q80 a1)

In this and similar passages Aquinas’ term ‘debere’ is suitably rendered by ‘ought’. It refers to the necessity that belongs to the requirements of the morally right. It is not imposed by any act of legislation.

Suarez recognizes the same difference between the requirement that we express through ‘ought’ and the sort of obligation that is imposed by an act of legislation:

For if this law forbids something because it is bad, it brings about its own special necessity of avoiding it, because this is intrinsic to forbidding. At the same time, however, it proves that this law assumes something which pertains to an intrinsic duty (debitum) of nature, because everything in a particular way owes it (debet) to itself to do nothing that conflicts with its own nature. But beyond this duty (debitum) the law adds a special moral obligation, and we say that this obligation is the effect of this law. (Suarez, De Legibus ii 9.4)²⁴

He argues against the identification of moral duty (debitum) with legal obligation (obligatio), by asserting that legislation is not the only source of duties and oughts.

If we now return to Anscombe’s historical claims, we recall that in her description of the law conception she asserts that ‘ought’ is ‘equated in the relevant contexts with “is obliged” or “is bound” or “is required to”, in the sense in which one can be obliged or bound by law, or something can be required by law’. This claim about equation is incompatible with the clear distinction drawn by both Aquinas and Suarez.

Many (including Aquinas and Suarez) who believe that some principles of morality are enjoined by divine legislation also believe that we are morally required to, and morally ought to, obey divine legislation. Their belief is not simply the belief that obedience to divine legislation is enjoined by divine legislation. They believe that some moral reason requires this obedience. These moralists take ‘morally ought’ to introduce a different concept from ‘obliged by divine legislation’. Similarly, anyone who disagrees with them, and denies that we morally ought to obey divine legislation, takes ‘morally ought’ to be distinct from ‘obliged by divine legislation’. By claiming that acceptance of the law conception requires the identification of the two concepts, Anscombe makes these arguments unintelligible.

Anscombe’s historical error undermines her claim about the modern use of the moral ‘ought’. Since pre-modern Christian believers in divine legislation do not identify the concepts ‘morally ought’ and ‘obliged by divine legislation’, we have no reason to believe that ‘morally ought’ has mere ‘mesmeric force’ when it is used by someone who does not believe in divine legislation. An attractive alternative hypothesis is that we mean by ‘morally ought’ what Aquinas and Suarez (for instance) mean by it. Both historical and philosophical arguments are needed to confirm this hypothesis, but at least it is far more plausible than anything that Anscombe asserts about the moral ‘ought’.

7 Is there a Moral Sense of ‘Ought’?

The argument I have just offered may appear to miss Anscombe’s point. Even if I have shown that Aquinas and Suarez use ‘ought’ in moral contexts without essential reference to divine legislation, it may be objected that these uses do not involve the moral sense of ‘ought’, because they do not treat ‘ought’ as giving an ‘absolute verdict’; they are really only using ‘ought’ in the non-moral sense that Anscombe takes to be legitimate and indispensable. This objection raises a further question about Anscombe’s argument. So far I have conceded that there is a moral sense of ‘ought’ that distinguishes some uses of ‘ought’ from the ordinary uses. But is there any such sense?

Suppose that both Beth and Clara tell Albert that he ought to take care of his health. He asks them why he ought to. Beth reminds him that he wants to be a snooker champion, and points out that he cannot achieve his ambition if he ruins his health. Clara tells him that his young children depend on him for their support and upbringing. We might describe this conversation on the assumption that Beth and Clara made the same initial assertion (‘You ought to take care of your health’). Albert has asked one question (‘Why ought I to take care of my health?’) and Beth and Clara have both answered that one question; Beth has cited prudential reasons and Clara has cited moral reasons. The fact that the reasons for the two ‘oughts’ are different does not show that the ‘oughts’ have different senses. If Clara’s judgment conveys an absolute verdict, and Beth’s does not, this difference may result from the character of the reasons that underlie the ‘ought’, not from the sense of the ‘ought’.

Anscombe rejects this univocal account of ‘ought’. In her view, ordinary uses of ‘ought’ are not open to Hume’s objection that we cannot derive ‘ought’ from ‘is’. Hume’s objection applies only to the moral ‘ought’, because it is not a real predicate, but a word of merely mesmeric force. If Hume’s objection is sound when it is directed against the moral ‘ought’, so understood, but unsound if it is directed against the ordinary ‘ought’, the ordinary ‘ought’ and the moral ‘ought’ have different senses.

We have no reason, then, to accept Anscombe’s view that the moral ‘ought’ introduces a distinct sense of ‘ought’ unless we already accept her prescriptive analysis of the moral ‘ought’. If we do not already accept her analyais, the fact that both prudential and moral reasons might be given in support of a question ‘Why ought I to take care of my health?’ suggests that the question is univocal, and that the moral ‘ought’ is distinctive because of the kinds of reasons that it offers, not because it has a distinct sense. But if there is no distinct moral sense of ‘ought’, Anscombe’s account of it is an attempted answer to a question that does not arise.

8 Is the Moral ‘Ought’ an Ordinary ‘Ought’?

One might still object that I have missed Anscombe’s main point. Even if we deny that ‘ought’ has a special moral sense, we might still believe that she must be right to distinguish the use of ‘ought’ in moral judgments from the ordinary use of ‘ought’ that we find in Aristotle. To see whether this objection is justified, we need to go back to her explication, as far as it goes, of the ordinary use.

Her explication is obscure at some points that matter for her distinction between the moral sense and the ordinary sense of ‘ought’. She suggests that sometimes the ordinary sense is functional. (See the passage on oiling a machine, quoted in Section 1 above). In this sense we can truly say we say that a man ought not to cheat, because we believe that such a man is unjust, and that an unjust man is a bad man. Anscombe suggests, then, that this sense of ‘ought’ can be explained with reference to good and bad. She may intend to explain Aristotle’s use of ‘ought’ (when he says, for instance, that the virtuous person acts ‘as one ought’) by saying that it refers to what is needed to achieve some end. But this gives a complete explanation only if we can explain the character of the end without saying that it is what one ought to pursue. If the end for the virtues is the appropriate or proper end for human beings, or the end that is worth pursuing for them, it is the end that they ought to pursue. This use of ‘ought’ cannot be explained in purely functional terms, with reference to some further end.

If, therefore, we cannot eliminate deontic terms from the analysis of the end of the Aristotelian virtues, we have not shown that Aristotle has no use for ‘oughts’ that are not purely functional. Hence there may be no ‘Aristotelian gap’ of the sort that Anscombe’s argument requires. (See Section 1.) This question about Aristotle is worth discussing, but I will not discuss it further.²⁵ I have mentioned it only to show that Anscombe’s distinction between the ordinary and the moral senses of ‘ought’ is questionable. She envisages giving up the use of ‘ought’ in the moral sense while keeping its use in the ordinary sense. But I have argued that we may be using ‘ought’ in the ordinary sense when we consider what we morally ought to do.

9 Is anything Wrong with the Moral ‘Ought’?

My comments on Anscombe’s main argument have been negative. I have argued that her initial question is based on a misunderstanding (since there is no special moral sense of ‘ought’ to be discussed), that she gives inconsistent answers to her question (that the moral ‘ought’ once had descriptive content, but now has purely prescriptive meaning, and yet the same concept survives),and that her arguments both for the purely prescriptive analysis and for the descriptive analysis are inadequate (because she gives no argument for the purely prescriptive analysis, and because her descriptive analysis, referring to obligations imposed by divine legislation rests on historical and philosophical errors). If these criticisms are justified, her argument gives us no reason to believe that there is any special difficulty about the moral ‘ought’, or that our use of it rests on unjustified assumptions. In particular, it is not the result of any assumptions about divine legislation that some people might reject. If, as some people suppose, virtue ethics is supposed to offer some alternative to theories that involve claims about what we morally ought to do, Anscombe’s argument gives us no reason to turn to virtue ethics.

But I do not want to end with these purely negative conclusions, or to suggest that nothing useful can be learned from Anscombe. When we consider her objections to the moral ‘ought’, and we see what is wrong with them, we find some reason to accept three conclusions: (1) Our use of the moral ‘ought’ is perfectly legitimate. (2) It does not reflect any sharp historical discontinuities. (3) We have good reason to believe that we and Aristotle share a common conception of morality that uses the moral ‘ought’.

These conclusions are sweeping and inadequately defended, and in these respects they are similar to Anscombe’s conclusions. But they are strengthened by the arguments that she offers for the contrary conclusions. While one might suppose that the broad historical scope of her paper, and the unqualified conclusions that she draws from very brief arguments, deprive her discussion of any value, this is not so. She encourages us to reflect on her claims, and to formulate some alternatives that have a better prospect of being fairly close to the truth.