I

In ‘Mistake’, Prichard's main argument against ideal utilitarianism is as follows:

Prichard gives two arguments for (1.3):

One problem with this argument is that the ideal utilitarian can deny (1.1). Moore, for example, holds that an ‘ought’ can be derived from a claim about goodness, either because ‘ought’ is defined in terms of goodness (Principia) or because it is a synthetic a priori truth that one ought to maximize the good (Ethics).Footnote ⁵ Rashdall seems to hold the Principia view; Laird clearly holds the Ethics view.Footnote ⁶ On either view, there is no need to apply ‘ought’ to states of affairs; that a state of affairs is best simply entails that one ought to bring it about, if one can.

The point is not just that ideal utilitarians can deny (1.1). (1.1) is highly counter-intuitive. Prichard does not intend (1.1) merely to rule out inferences from the non-normative to the normative. It is to support thinking that the badness of pain is not ‘the reason why we ought not to inflict pain on another’ or that ‘we do not come to appreciate an obligation. . .by an argument of which a premiss is the. . .activity of appreciating the goodness. . .of a consequence of the act’ (M 10 n. 4, 13–14). Strangely, Prichard concedes ‘that unless we recognized that something which an act will originate is good, we should not recognize that we ought to do the action’ (M 10). The ‘goodness of the thing produced is a presupposition of the obligation to produce it’ (B 2–3). He adds that ‘this is not to imply that. . .the badness of pain [is] the reason why we ought not to inflict it without special cause’ (M 10).Footnote ⁷ He is right that the implication does not follow; but the obvious explanation for why producing goodness is a necessary condition on an ‘ought’ claim is that suggested by the ideal utilitarians.

Prichard gives no argument for (1.1). He does cite Rashdall's claim that ‘[t]he good is that which “ought to be”’ (MO 210, 224; also M 9).Footnote ⁸ (Moore also writes of the good as what ‘ought to exist’, ‘ought to be’, ‘what we ought to aim at’, what ‘ought to be [one's] end’, or that which ‘we ought to try to bring about’.Footnote ⁹ ) Prichard may think Rashdall concedes (1.1). Rashdall's view, however, is not so clear (see Appendix).

A different problem is that some ideal utilitarians might accept (1.1) – holding ‘one ought to maximize the good’ as underived. They have no need to say that the good ‘ought to be’, since they are not deriving ‘ought’ claims from claims about goodness at all. Perhaps Prichard does not intend to target this view.Footnote ¹⁰ But if so, his argument misses one sort of ideal utilitarianism. (Prichard could, however, ask why this is the only duty.)

II

In ‘Moral Obligation’, Prichard argues that ideal utilitarians are guilty of ‘resolving obligation into something which it is not’ (MO 225). They make ‘a mistake which is far more radical than any which we should even be inclined to suspect beforehand’ (MO 221).

Here (2.2) is the questionable premiss. (2.2) rules out any inference from properties of the act to obligations of agents. As above, it rules out the inference from the fact that an act has the property of causing pain to an obligation not to perform that act. But (as Prichard notes in correspondence) it rules out a great deal more; it also rules out the view, held by Ross, Carritt, Broad and indeed the earlier Prichard, that one can infer from the nature of the act to obligations of agents.Footnote ¹² For example, earlier Prichard wrote that ‘[w]e recognize. . . that this performance of a service to X, who has done us a service, just in virtue of its being the performance of a service to one who has rendered a service to the would-be agent, ought to be done by us’. Indeed, Prichard thought this fact is self-evident. Similarly, he noted that it is wrong to suggest that ‘even if the nature of the act is completely stated, it is still necessary to give a reason’ (M 13) and that ‘in thinking of our keeping our promise to X as a duty, we are thinking of the action as rendered a duty by its being the keeping of our promise’ (DI 29; also B 4). The point of moral principles is that they bring out ‘just that feature of the act which constitutes it an obligation’ (B 5).

Prichard does not argue for (2.2), but he does address Joseph's denial of it. Joseph writes that ‘[a]n obligatory act is like a well-remembered face; the face no doubt has characters because of which it is well remembered, but it is called well-remembered to signify not those characters, but that others remember it’.Footnote ¹³ A property of us – that we well remember the face – arises from a property of the face (a hooked nose, say), just as a property of us – that we are obligated – arises from a property of the act. Prichard replies that ‘it cannot be maintained that what renders a face well remembered is its having certain distinctive features, as distinct from our having noticed that it has them, without implying that what is called its being well remembered is a certain character of the face’ (MO 222–3). Prichard's point might be that something more than a hooked nose is necessary to make a face well remembered. I must, in addition, notice the nose. A property of the face does not, on its own, give rise to a property of us. But this defence of (2.2) supposes that the ideal utilitarian must hold that properties of acts, on their own, give rise to properties of agents. The ideal utilitarian need not require this. A property of an action – that it maximizes the good – can give rise to a property of us – that we are obligated to do it – given that we have an obligation to maximize the good. (Similarly, the hooked nose can give rise to our remembering it, given that we notice it.)Footnote ¹⁴

III

After giving this argument, Prichard considers what can be said against one who rejects (2.4), the view that obligations are properties of us rather than actions.

In his correspondence, Prichard also uses the thought behind this argument to support a restricted version of (2.2): ‘for the character of the action to necessitate anything, the action has to be actual, i.e. to exist, so. . .there could only be obligations to acts which we are actually doing’.Footnote ¹⁵ A property of the action cannot give rise to a property of the agent, because the act need not exist when the agent has the property. An obligation based on properties of the act is ‘like a cat's grin without the cat’.Footnote ¹⁶

One reply, made by Joseph, Broad, Ross, and Carritt, is that properties the act would have, were it performed, can make me obligated.Footnote ¹⁷ For example, if an act would cause great and pointless misery, that is a reason for me not to perform it. It is hard to see how Prichard himself can avoid this inference from properties an act would have to obligations of agents. Earlier he wrote that ‘obligation admits of degrees, and that where obligations conflict, the decision of what we ought to do turns not on the question “Which of the alternative courses of action will originate the greater good?” but on the question “Which is the greater obligation?”’ (M 14 n.). It is hard to see how I can determine the greater obligation without considering properties of rival possible actions such as ‘that would fulfil a trivial promise’ or ‘that would prevent me from keeping an important promise’. Or again, Prichard wrote that ‘we may not appreciate the wrongness of telling a certain story until we realize that we should thereby be hurting the feelings of one of our audience’ (M 12).Footnote ¹⁸ This looks like an inference from a property the act would have, if performed, to an obligation not to perform it.

Another reply, suggested by Broad, is to find something that exists before the act is performed, such as a description of the act, to which properties can be attributed. Broad suggests that ‘x ought to exist’ asserts of a description of x that it ought to have an instance.Footnote ¹⁹

Before leaving this argument, it might help to note a statement of it earlier in ‘Moral Obligation’. In saying that ‘X ought to educate Y’ we cannot be attributing a property to the act, because ‘we can no more either think or assert of something which does not exist that it ought to exist than we can think or assert anything else about it. Of what we think does not exist we can think and assert nothing at all.’ Prichard then suggests that we could mean ‘“[i]f X is or will be educating Y, his educating Y is, or will be, something which ought to exist.”’ He allows that what is grammatically categorical – ‘X ought to educate Y’ – might be hypothetical in meaning. But he dismisses the proposal: ‘we find ourselves forced to admit that when we assert, e.g., that X ought to be educating Y, we are attributing a certain attribute to X, and therefore are implying the existence of such an attribute as the being under an obligation to do some action’ (MO 168–9).

Two things stand out. First, Prichard does not reject speaking of properties an act would have, were it performed. He does not really mean that we can ‘think and assert nothing at all’ about non-existent things.

Second, Prichard considers ‘if X educates Y, his educating Y will be something that ought to exist’ as giving the meaning of ‘X ought to educate Y’. He then rejects it as not giving this meaning, since in saying ‘X ought to educate Y’ we are attributing a property to X. He does not consider that ‘if X educates Y, his educating Y will be something that ought to exist’ might be offered as giving the ground of the obligation rather than the meaning of ‘X ought to educate Y’. This would let one think both that X has the property of being under a moral obligation and that the obligation depends on properties of the possible acts. (In correspondence he does consider this, but his response is merely the unsupported claim that hypothetical facts cannot ground categorical ones.Footnote ²⁰ )

IV

In ‘Moral Obligation’, before giving the arguments I have considered in II and III, Prichard gives arguments against what he calls ‘teleological’ and ‘quasi-teleological’ theories of obligation. ‘Teleological’ theories explain obligation by aims the agent possesses; ‘quasi-teleological’ theories explain obligation by aims the agent ought to possess. For Prichard, it seems that a theory is not teleological (of either sort) in virtue of basing obligation on the production of the good, but rather by basing obligation on the aims of the agent. Thus when Prichard interprets Sidgwick as basing obligation on what we ought to bring about rather than on what we ought to aim at, he notes that Sidgwick is refuted not by the arguments against quasi-teleological theories, but by those I consider in II, III and VII, against basing obligation on goodness (MO 207 n. 58).

Prichard distinguishes between two sorts of ‘ought’: there are moral ‘oughts’ found in categorical imperatives, and non-moral ‘oughts’ found in hypothetical imperatives. He thinks that the ‘oughts’ found in hypothetical imperatives are not really ought-claims at all, since they can be replaced by saying that a certain means is efficient for attaining a certain end (MO 166, MM 126–8, DI 34, 43, K 54–5).Footnote ²¹ But he could give his argument without thinking this, as long as the ‘ought’ of hypothetical imperatives is not the ‘ought’ of moral obligation. His objection is that utilitarians justify only non-moral ought claims: they take one's aim to be to maximize happiness, and conclude that, say, one ought to tell the truth because that will maximize happiness; here the ‘ought’ is the ‘ought’ of a hypothetical imperative (MM 135, 142–5).

Prichard does not direct this objection against the (obviously preferable) quasi-teleological interpretation of utilitarianism, on which utilitarians say one's aim ought to be to maximize happiness, or against the (even more preferable) interpretation on which utilitarians say that one ought to bring about the maximum happiness. Against these views, he could still charge that there is no obligation to tell the truth, since the ‘ought’ remains hypothetical. But he could not charge that utilitarianism ‘can be ruled out on the general ground that it substitutes the non-moral for the moral “ought,” and is therefore not a view of morality at all’ (MM 145; also 143, 144). For the ‘ought’ in ‘one ought to aim at the general happiness’ or ‘one ought to bring about the general happiness’ is not the hypothetical ‘ought’. Prichard must, then, rule out the ‘ought to aim’ and ‘ought to bring about’ views.Footnote ²²

He has two arguments against the ‘ought to aim’ view. First,

One reply is to deny (4.4) and (4.5). Broad notes that we say ‘“You ought to be ashamed of yourself”’ or ‘“You ought not to wish for another's humiliation.”’Footnote ²³ Broad seems to be taking the ‘ought’ in the ideal-stating sense which does not imply ‘can’, but perhaps this is how a quasi-teleologist should understand ‘ought to aim’ in the first place.Footnote ²⁴

The second argument against the ‘ought to aim’ view would let Prichard concede Broad's point. The objection is that one cannot argue from claims about what we ought to desire to claims about what we are obligated to do.

Where A is the means to X, it does not follow that I ought to do A unless I actually aim at X. That I ought to aim at X is irrelevant.

Whether this is true is very dubious. If, say, I grant that a better-informed version of myself would advise me to have X as an end, it seems very plausible to think that I ought to do what conduces to X. What is irrelevant here is that I actually aim at something other than X. There is an ‘ought’ that, unlike Prichard's non-moral ‘ought’, does not depend on my actual aims, but like Prichard's non-moral ‘ought’, does transmit normative force from ends to means. Prichard himself suggests, without criticism, something close to this: ‘we, being only imperfect, do not always have the existence of what is good as our purpose, yet we can only find out what in a moral sense we ought to do by working out what would be the actions we ought to do in the non-moral sense, if the existence of what is good were our purpose’ (MM 146).

Prichard might have replies to these objections to his arguments against the ‘ought to aim’ view. But even if he does, the big problem is that the utilitarian can escape by holding the ‘ought to bring about’ view and say nothing about aims.

Prichard considers this. He notes that although Sidgwick sometimes writes of aiming, his real view may be that ‘where something would be good this fact involves that we ought to do what would most conduce to it, the notion of “end” or “purpose” having nothing to do with obligation’ (MO 207 n. 58; also MM 135, 136–7). In ‘Moral Obligation’, Prichard then refers the reader to the objections I consider in II, III, and VI (MO 207 n. 58).Footnote ²⁵ In ‘Manuscript’, instead of doing this, Prichard then objects that Sidgwick comes out as holding an ‘intuitional’ view, since ‘if he had been asked “how according to common sense do we come to apprehend that such actions [as making others happy] are right?” he could only have answered that, according to common sense, we do so directly’ (MM 137).Footnote ²⁶ But as Prichard admits later, Sidgwick could say, as Prichard has Moore and Rashdall say, that we come to apprehend that it is right to make others happy by seeing that happiness is good (MM 151–2; 152–3). The crucial arguments, then, are not those against the teleologists and quasi-teleologists, but against those who base obligation on the good.Footnote ²⁷

V

In ‘Mistake’, after giving the argument considered in I, Prichard adds that

This objection was made by Butler and Price; it becomes the centrepiece of the second chapter of Ross's The Right and the Good. Here I note two features of Prichard's statement of it.

(i) One version of the objection claims that the ideal utilitarian gives the wrong verdict. Prichard claims this for justice. When Butler and Price made this objection, their target was hedonistic utilitarianism. They did not need to worry that the utilitarian might, by expanding what counts as good, get the right verdict – by, for example, counting just distribution as itself a good.Footnote ²⁸ Although Prichard takes as his target ‘Utilitarianism in the generic sense, in which what is good is not limited to pleasure’, he does not consider this reply (M 9). Carritt, and Ross by the time of Foundations, do: they argue that while the utilitarian can add goods to get the right verdicts, the justification for thinking of these states of affairs as good is a prior judgement that one has a duty to, for example, pay one's debts.Footnote ²⁹

Prichard may not have considered the expanding-the-good tactic for two reasons. First, he seems to have a very restricted view of what is good. In ‘Mistake’, he allows that knowledge is good and pain bad (M 10). But later he argues that this is false – only capacities such as virtue are good (MO 173).Footnote ³⁰ This would exclude the states of affairs employed by the ideal utilitarian. In addition, his view that actions are good only in virtue of their motives (M 11, B 3, MM 155–6) would rule out thinking that, say, that an action is an instance of promise-keeping makes the action good. But by itself it does not rule out thinking that the result of the action – that a promise has been kept – is good.

Second, Prichard may not have considered the expanding-the-good tactic because of his view of consequences. Prichard writes that ‘[b]y “a consequence of the act”. . . we mean a particular something Y caused by X, and so brought about by the agent indirectly’ (MM 139; also 141). If ideal utilitarianism is stated as the view that an action is made right by its good consequences, this view would exclude counting the state of affairs that-the-action-has-been-performed as a good consequence.Footnote ³¹ The same view underlies Prichard's claim that Moore and Rashdall are inconsistent for thinking that both the good results of an action and the goodness of the action itself can make an action right (MO 212–13, MM 152; also MO 209). But this view of consequences and results yields an uninteresting semantic victory. Well before Prichard wrote, Rashdall had included the value of both acts and their consequences, and Broad had argued that ideal utilitarians should take ‘the total state of the universe’, ‘past, present, and future’, as fixing rightness.Footnote ³²

(ii) Another version of the objection claims that the ideal utilitarian may get the right verdict, but gives the wrong reason for it (see also MM 136).Footnote ³³ In Prichard's example, the ideal utilitarian thinks he should pay his debt to A because doing so will bring about A's material comfort. The worry might be that we think only that we owe a debt, rather than of what A might do with what is owed. But the ideal utilitarian need not cite consequences that far down the line. If the good is expanded to include the state of affairs in which debts are paid, the ideal utilitarian can think that he should pay his debt to A because doing so will bring it about that the debt has been paid. In the case of lying and true belief, the ideal utilitarian might similarly say that he should not lie because the state of affairs in which lies are told is bad – or perhaps here it does not seem wrong to cite the true belief in B as the reason for not lying.

Prichard might reply that the problem is not that the ideal utilitarian cites consequences too far removed from those we really cite. It is rather that in deciding upon our duty, we do not think of these states of affairs as goods at all. But while this may be sometimes, or even often, true, the ideal utilitarian can note that if, on reflection, we do think these states of affairs are goods, it does not matter much that we do not in every instance of acting have this thought.Footnote ³⁴

VI

In a footnote in ‘Mistake’, Prichard writes that

Some might describe it as wrong. But the more important reply is that the ideal utilitarian's claim that the badness of pain gives us reason not to inflict it on ourselves or others is unchallenged. That we have reason not to inflict it on ourselves explains why it would be foolish to do so.Footnote ³⁵ We may not speak of ‘wrongness’ here, but that might be because we reserve ‘wrong’ for actions we are most concerned to prevent, or even because many have come to think that morality is silent when my actions do not harm others.Footnote ³⁶ Even if we do not call the reasons in our own case ‘moral reasons’, this can be seen as an irrelevant fact about language, rather than a reflection of what the badness of pain can explain.

VII

In ‘What is the Basis of Moral Obligation?’ the following argument replaces the argument, criticized in I, that turns on speaking of what ‘ought to be’. Prichard asks one to suppose that

Here Prichard makes two now-familiar objections to the impartiality of ideal utilitarianism: the man is more obligated to help his father than the stranger, even if equal amounts of good are produced; and (as in the ‘Mistake’ footnote) there is a self–other asymmetry, in that I am more obligated to suppress my mean thoughts than those of another, even if I could do either with equal ease, and the man is permitted to help others rather than himself, even if less good is produced.Footnote ³⁸

Prichard writes ‘not merely because the latter is more difficult’ to rule out one of the familiar ideal utilitarian explanations of the difference in duties.Footnote ³⁹ There is, however, another familiar reply Prichard does not rule out: again the good might be expanded, now to include (say) the state of affairs in which sons help fathers. Of course there is a familiar counter-reply: if I can either help my father, or perform some act that would enable five other sons to help each of their fathers, the ideal utilitarian would tell me not to help my father. Some now take this sort of example to be decisive against ideal utilitarianism. Thomas Baldwin, arguing against Moore, says there is ‘[s]urely not’ a good reason to perform the act that would enable five others to help each of their fathers. This is ‘surely uncontentious’.Footnote ⁴⁰ Judith Thomson has the same view.Footnote ⁴¹ Perhaps Prichard simply did not think of this sort of example. But there may be another explanation.

It is striking that the arguments I have considered in this section and the previous two – arguments that dominate the debate now – get so little attention from later Prichard. Each barely appears in the longer, later works. The claims Prichard came to rely on (and in the first case relied on even in ‘Mistake’) are principles like ‘an “ought” can be derived only from another “ought”’ or ‘“ought” refers to actions alone’ or ‘the property of an action can give rise only to another property of that action’ or ‘we cannot think or assert anything of things that do not exist’ or ‘hypothetical facts cannot ground categorical facts’. These are the claims – often about what derivations are permissible between types of statements – about which Prichard is most confident.Footnote ⁴²

This point might correct one view of Prichard. The end of ‘Mistake’ is naturally read as suggesting that Prichard has knowledge of, and is dogmatic about, moral judgements. We should ‘realize the self-evidence of our obligations’. Doubts can be allayed by ‘getting face to face with a particular instance of the situation. . .and then directly appreciating the obligation’ (M 20). But earlier he notes that ‘owing to a lack of thoughtfulness, even the best men are blind to many of their obligations’ (M 14 n.; also B 6). ‘Basis’ concludes with Prichard claiming that ‘owing to the complication of human relations, the problem of what one ought to do, from the point of life as a whole, must present itself to any thinking man as one of intense difficulty and, indeed, doubt’ (B 6).Footnote ⁴³ Perhaps this is why, unlike both Ross and recent critics of ideal utilitarianism, he came to rely more on the abstract principles rather than the moral judgements. In one way, this is a good thing – I think the intuition Baldwin and Thomson rely on is not shared as widely as they believe.Footnote ⁴⁴ But since I have argued that the abstract principles do not fare so well either, this may offer Prichard small consolation.