"The Doctrine of Consequences in Ethics," International Journal of Ethics 24 (1914): 293-320. Reprinted in Broad's Critical Essays in Moral Philosophy,ed. David R. Cheney (New York: Humanities Press, 1971).
THE DOCTRINE OF CONSEQUENCES IN ETHICS
C. D. BROAD
The opinion that the rightness of an act is in some way connected with the goodness or badness of its consequences is, I suppose, held by everyone in practice and by most moralists in theory. If we only listen to what people say instead of also noticing how they act and judge, we might be inclined to underrate the amount of agreement on this point. Nothing is commoner than such phrases as 'you must never do evil that good may come,' which, if they mean anything, imply that some acts are wrong, however good their consequences. Yet, in practice, people who quote this maxim and also believe that pain is an evil do not, as they ought to do, shun their dentists as moral lepers. Again, there is no doubt that commonsense thinks motives important as well as consequences, but it would reject the Kantian view that they are all-important, and that only one kind of motive is morally valuable.
But at this point agreement ceases. Are consequences the sole relevant factor in judging the rightness of an act; or do other factors enter, and, in particular, are some acts right and some wrong, whatever their consequences? Again, is it the actual or the probable consequences that are ethically relevant? And further, if you decide to include motives in judging the rightness of an act, is the question whether the act is the immediate response of a good nature or results only as the consequence of a moral struggle, of ethical importance? The first two of these questions have been discussed with considerable fulness in recent years by Mr. Russell in his Philosophical Essays, and by Mr. Moore in his little book on Ethics. The names of these two philosophers are a sufficient guarantee for the ingenuity and subtlety of their arguments; but, since they disagree, one must be wrong, and of course both may be. It will, therefore, be of some interest to take their views as a text for the discussion of the subject of this paper.
I will begin with the points on which Moore and Russell are at one. Both seem to hold that you can only talk of right actions, and not of good ones. "What is called good conduct," Mr. Russell says, "is conduct that is good as a means to other things that are good on their own account." If this be the whole of what is meant by good conduct, it is no doubt well to do as Mr. Russell proposes and call it 'right' and its consequences alone 'good,' because otherwise 'good' is used in two different senses when we speak of 'good conduct' and of 'good consequences.' Now I cannot help thinking that neither of our authors has given enough attention to the possibility that there may be good conduct in their sense of good, as well as right conduct in their sense of right, and that the two need not coincide. This possibility arises in two different ways. In the first place, if other states of mind be intrinsically good or bad, I do not see why volitions should not have intrinsic goodness or badness. I should suppose that most people think that they have, and the question is at least worth discussion. Moore and Russell scarcely touch on this point, but I conclude that their opinion is that they are intrinsically indifferent. Mr. Moore says (Ethics, p. 185): "It is contrary to our view that motives can be intrinsically good or bad. But, if it is true, it makes no difference to the rightness or wrongness of an action, but only to the goodness of a total state of affairs." The latter part of the statement is unquestionably true on Mr. Moore's definition of rightness, because that explicitly only refers to consequences and therefore does not include the motive or volition that precedes them, so that the intrinsic value of the latter does not enter into consideration. But this only shows that Moore is consistent with his own definition, not that he means by rightness what other people mean when they understand themselves. But this, I suppose, is what he is trying to show.
We have seen then that it is at least possible that some volition may be intrinsically good, and therefore, since an act is certainly not identical with its own consequences, there may be good acts as well as right ones. But even if Moore be right in thinking that no motive is intrinsically good, it will not by any means follow that the intrinsic value of the consequences x + the volition a will be the same as that of the consequences x + the volition b. For to assert that because a and b are intrinsically indifferent, therefore ax and bx must necessarily have the same intrinsic value would be to forget the Principle of Organic Unities. I do not suppose that Moore or Russell would really deny anything that I have been saying; all that I suggest is that it is of some importance and seems to have been overlooked by them. I do not quite know whether Mr. Russell thinks that anything but consequences are good, for he makes the curious statement (Philosophical Essays, p. 31): "I do not wish to deny that right conduct is among the things that are good on their own account, but, if so, it depends for its intrinsic goodness upon the goodness of those other things which it aims at producing." As it stands, this is surely contradictory; a thing cannot depend for its intrinsic goodness on anything else. What I suppose Russell must mean is that it is not intrinsically good, but, when added to its consequences, may produce a whole whose intrinsic value differs from that of the whole formed by the consequences only.
So much then for the agreement between Moore and Russell. As some good consequences are certainly states of mind, we shall be inclined to suppose that the states of mind that result in consequences may also have intrinsic value; and, even if they have not, the Principle of Organic Unities forbids us to deny the possibility of good conduct differing from right conduct in Moore's sense of the word. I shall return to this question at a much later stage of the paper.
Let us now consider the differences between the two writers. They differ about rightness. Russell has a complicated theory which introduces the probable consequences of actions. Moore's theory is simpler, and, at first sight, much more paradoxical since it makes rightness depend wholly on the actual results of our actions. It further makes much use of the notions of justifiable praise and blame. As a matter of fact, we shall see that really both theories have to introduce probability, and that Russel's view when worked out is much less plausible than it seems at first sight. I will begin by sketching and criticising Russell's theory.
That act which has as good consequences1 as any that is open to the agent is called by Russell 'a most fortunate act,' and by Moore a 'right act.' But Russell will not admit that a most fortunate act is necessarily a right one. He grants that there is an objective and a subjective sense of right, but holds that even the objective sense is partly subjective, whilst what is most fortunate is quite independent of our knowledge and belief. If the evidence be against an actually most fortunate act being most fortunate, it is objectively wrong to go against the evidence. An objectively right act is what he calls a 'wisest act,' i.e., one which is probably a most fortunate act. This theory, he says, has the advantage that it makes unforeseen factors irrelevant to rightness and wrongness. Now objectively, of course, a man ought to do what is objectively right; but there is another sense of ought, Russell holds, in which we must say that he sometimes ought to do what is objectively wrong. This second sense of ought, of course, involves another sense of right. A man acts rightly in this sense (or conscientiously, as we may call it) if he does what he judges to be objectively right, after he has reflected with a view to finding out the truth on this point, and not merely with a view to proving one course right. If the act be unimportant, and if it would need much reflection to come to any decision about its objective rightness, it is morally indifferent.
We may admit that this is an ingenious and plausible theory which seems to cover most of the obvious facts. But I think we shall find that it is not nearly so simple as it looks. It is not explicitly remarked by either Moore or Russell that there is a very close analogy between the three meanings of ought in ethics which between them they use, and three meanings of ought in logic. I believe it will be helpful to develop this analogy a little. When I ask: 'What ought I to believe?' one answer certainly would seem to be: 'What is true.' Now, 'what is true' corresponds here to a most fortunate act in ethics in its complete independence of anyone's knowledge or belief. So this answer corresponds to that which Moore makes to the question: 'What ought I to do?' But there certainly seem to be plenty of cases where in a sense I ought to believe what is actually false. If I believe that all M is P and that all S is M, there is certainly a sense in which I ought to believe that all S is P (or, at any rate, ought not to believe that some S is not P). Yet if one of the first two beliefs be false, it may very well be the case that what I ought to believe about the relation of S and P is false, and what I ought not to believe is true. Nor is it relevant to answer: You ought not to have believed, e. g., that all M is P, since that is false. Even if this be so (and to assert it is to beg the question as to whether there are not several meanings of ought in logic), yet we must still ask: What ought I to believe, granted that as a matter of fact I have this belief, which, of course, I do not know to be false so long as I have it! The answer: You ought to believe, or perhaps I should rather say, you ought not to disbelieve what logically follows from what you do believe, corresponds rather closely with Russell's objective sense of ought and right. Then, finally, it might be suggested that you ought to believe what you think logically follows from what you do believe. If you really think that A propositions can be simply converted and believe that all S is P, then you ought not to believe that some P is not S. This seems to correspond to subjective rightness in ethics.
I think, then, we may fairly suggest that Russell's theory of the different meanings of right in ethics can probably be reduced to considerations involving the different meanings of ought in logic + what seems to be a purely ethical meaning of ought which appears equally in both his senses. The ethical common meaning is involved in the statement: I ought ethically always to do that action which I ought logically to judge a most fortunate action of those possible to me. Russell's two meanings of ought are syntheses of this common third ethical meaning with the two logical senses of the word. Russell's qualification about reflecting with a view to finding out what is objectively right is involved in the logical sense of ought; for you have no right in any logical sense to believe what is not self-evident and what you have not investigated with a desire to reach the truth, whatever it may prove to be.
What is objectively right, then, on this theory is what, on my information, has probably at least as good consequences as any other action open to me. The mixture of the objective and subjective here is nothing specially ethical, but is what is involved in all applications of probability. The point is simply this. Any proposition, whether about goodness or anything else, is either true or false. But, relative to various selections of other
propositions, this proposition will have different degrees of probability. So far, all is perfectly objective; the probability of one proposition, relative to any definite selection of others, is as independent of anyone's knowledge or belief about it as its truth or falsehood. When the individual mind enters, it is simply as a selective agent. Any particular mind believes some proposition and not others; and is acquainted with some and not with others; and relative to those which it believes, any given proposition has a certain probability, whether this particular mind or any other knows its value or not. It is necessary to add, however, that, whilst there is no ambiguity, and nothing subjective, in the probability of a proposition relative to any definite selection of others, there is some vagueness as to what Russell supposes to be included in the selection involved in his theory of objective rightness. This is really an important matter. Does objective rightness depend on the probability of the consequences relative to all the propositions that the agent believes; or to all those that he believes and the contradictories of those that he disbelieves; or to all the true propositions that he believes and the contradictories of all the false ones that he disbelieves; or what precisely is the principle of selection? It seems to me that Russell has talked cheerfully about the probabilities of propositions without remembering that all probabilities are relative to selections of propositions, and that it is vital to state what selection he considers relevant in defining rightness. When this fact is taken into account, the distinction between objective and subjective rightness on Russell's theory becomes a somewhat subtle one, though I grant that it remains. My doubt is whether it corresponds to any distinction involved in the various ethical judgments of commonsense which led Russell to his theory of objective rightness. Let me elaborate this point a little.
On Russell's theory it is clear that of precisely similar acts performed by A and B, under circumstances agreeing in all respects but in the knowledge or beliefs of the agents, one can be objectively right and the other objectively wrong. For, since the probability of the consequences is relative to different selections of propositions in the two cases, it will in general be different. Now, of course, we all admit that difference of knowledge may alter the subjective rightness of acts performed under otherwise precisely similar circumstances; but I do very much doubt whether any judgment of common sense implies that what is objectively right for A can be objectively wrong for B. Still I do not wish dogmatically to assert that there is not an ethical sense of right and ought different from what Russell calls subjectively right and from what he calls most fortunate. My lingering doubt in his favor is due to the fact that there does seem to me to be such an intermediate meaning of right and ought in the logical sense, and that Russell's objective rightness seems, as I have said, to correspond to this.
I said that the logical sense of ought that corresponds to Russell's objective rightness is that you ought to believe or at least not to disbelieve what logically follows from what you do believe and from the contradictories of what you disbelieve. It is clear that this does not imply that what you believe is true, or what you disbelieve false; for otherwise this kind of logical rightness would not be (as it certainly is) compatible with its being right to believe what is false, and to disbelieve what is true. Now the probability of one proposition relative to others seems to me to be as much a matter of pure logic as its being implied by them. Hence, we must include in this logical sense of right that it is right to attribute to any proposition that degree of probability that it actually has, relative to the propositions that you believe and to the contradictories of those that you disbelieve. If my analogy between this logical sense of ought and Russell's objective rightness be accepted, we have now made a beginning of answering the question, -- which, as we saw, he left unanswered, -- as to what precisely is supposed to be the selection of propositions; relative to which the probability involved in objective rightness is to be reckoned. The relevant propositions are roughly those which the agent believes and the contradictories of those which he disbelieves, irrespective of whether what he believes is true or what he disbelieves is false. But some modification must be made in this.
The distinction between the use of a proposition as a premise and as a principle is familiar to logicians. When I argue in accordance with Barbara and justify my procedure by pointing this out, I do not use Barbara as a premise, but as a principle. Now, for logical rightness we must qualify what I have already said by adding that it is only the truth or falsity of a man's beliefs in propositions which he uses as premises that is irrelevant to his logical rightness; we must assume that his beliefs in all propositions that he uses as principles are actually true. This is, of course, simple enough when we are dealing with certain inference; but it is less easy to see what is a premise and what a principle when we come to deal with probability. In the first place, it is clear that the purely formal laws connecting the probability of a complex proposition, conjunctive or disjunctive, with that of its separate elements must be taken as principles and not as premises. I must also never count among the premises, relative to which I reckon the probability of a proposition, propositions about its probability relative to the other premises. Such propositions are principles not premises, and my belief in them must be true if I am to be logically right. For, otherwise, the true probability of anything relative to my state of information, would depend on my belief about the true probability, and this would make the whole notion purely subjective, which it is not. And this is supported by the analogy with certain inference, since to be logically right there, I must not have false beliefs as to whether one proposition implies another or not, and the case of implication is one where the probability of what is implied relative to what implies it is 1.
We now see what is the logical sense of rightness that we take to correspond with Russell's objective ethical rightness. Does his ethical objective rightness involve anything further than that it is objectively right to do what it is logically right to believe to be a most fortunate action? There are great difficulties and ambiguities in the notion of a probably most fortunate action which I will deal with later, but at present the question is where precisely the ethical element enters, and whether it introduces any new question of principle. It is, of course, clear that to reach judgments, about the probable goodness of anything, there must be some premises about values believed, as well as premises about facts. Can these ethical premises be treated as precisely on a level with the factual ones or not; i. e., is it only the question whether we believe or disbelieve the ethical premises that is relevant to objective rightness, or is the truth or falsehood of our beliefs and disbeliefs here of importance? I am inclined to think that there is a difference between the two kinds of premises in this respect.
Suppose, for instance, that a person is an ethical hedonist, i. e. that he believes as one of his ethical premises that the goodness of any state of affairs is directly proportional to the amount of pleasure in it and to that alone. Relative to this proposition, an act that will probably produce more pleasure on the whole will probably have better consequences than one that will probably produce less pleasure. But, supposing that ethical hedonism is false, are we to say that the man's act is objectively right, if he is right about the probability of its pleasurable consequences? If, for example, one is logically justified by one's factual premises in holding that pushpin will probably give more pleasure on the whole than poetry, and if one is logically justified on one's ethical premises in holding that more pleasurable states are always better than less pleasurable ones, is it objectively right to prefer pushpin to poetry, even though your ethical belief be false? I do not think this can be maintained, and therefore there must be an important distinction between the positions of ethical and of factual beliefs in the matter of objective rightness. Let us, then, sum up the results of our attempts to clarify Russell's notion of objective rightness, as far as they have yet appeared. An act is objectively right if it is probably a most fortunate act relative to
- the propositions about matters of fact which the agent believes, and the contradictories of those which he disbelieves, independently of whether they be true or false;
- to true ethical premises, whether he believes them or not; and
- to true principles of inference and probability, whether he believes them or not.
Subjective rightness depends wholly on what people believe or disbelieve, and not at all on what is true, while Russell's objective rightness depends, partly, on what people believe, whether it be true or not, and, partly, on what is true, whether people believe it or not.
But we are by no means at the end of our difficulties. Russell does not seem to have remarked that the notion of a probably most fortunate act remains ambiguous even after you have defined the selection of propositions, relative to which its probability and that of alternative acts are to be reckoned. The fact is that the question of probability enters twice, and Russell has not distinguished its two appearances. It is not clear whether the objective rightness of an act depends on the actual value of its probable consequences, or the probable value of its actual consequences, or the probable value of its probable consequences. All that we are told is that it does not depend on the actual value of its actual consequences. I submit that until the theory that objective rightness depends on probable consequences decides between these three alternatives, it cannot be satisfactory, and that when it is faced by them, it loses some of its original plausibility.
We must devote a moment to the consideration of these alternatives. So far as I can see, the most plausible view for Russell to take would be that rightness depends on the probable value of probable consequences. For if either the actual goodness or the actual consequences be relevant, it is difficult to see why both should not be; and this he denies. If, however, we work out the implications of this theory, we shall see that it is less simple than it looks. Suppose that an agent has two actions, X and Y, open to him. Suppose, further, that relative to the propositions that the agent believes and disbelieves, the most probable consequences of X are A, and that their probability is p. Let the most probable consequences of Y be B, and let their probability be q. Further, let the most probable measure of the goodness of A be x, and the most probable measure of the goodness of B be y. Now suppose that p < q and x > y. What then is objectively right? Ought the man to choose the act whose most probable consequences are less probable, but most probably more good, or the one whose most probable consequence is more probable, but most probably less good? It is useless to say that the question is merely academic, since the calculations cannot be made, for it is quite irrelevant to objective rightness whether anyone actually makes the calculations or not. The difficulty is one of principle, and, unless the theory can remove it, it has not produced an unambiguous definition of what it means by objective rightness.
Of the two remaining alternatives, it seems to me that it is more plausible to suggest that objective rightness depends on the actual value of probable consequences than on the probable value of actual consequences. It is, in fact, clear that the latter is not what Russell means, since he congratulates his theory on making objective rightness independent of unforeseeable circumstances, i.e. of true propositions about matters of fact which are not included in the selection believed by the agent at the time of decision. Let us, then, take the view that an act is objectively right whose most probable consequences would be actually at least as good as the most probable consequences of any other act open to the agent. Unfortunately, there is much the same ambiguity here as we noticed above. If I can perform either X or Y, and the most probable consequence of X is A and of Y is B, it will not, in general, be true that the probability of X being followed by A is the same as that of Y being followed by B. If the probability of the most probable consequence of X be p, and that of the most probable consequence of Y be q, when p > q, whilst the value A is x and that of B is y where x < y, which act is objectively right? We could only avoid this ambiguity by introducing the notion of expectation, i. e., the product of the probability of an event by the actual or most probable measure of its goodness if it takes place. We might then say that an act is objectively right if the expectation of goodness is, relative to the selection of propositions already defined, at least as great as the expectation of goodness of any other act possible to the agent. But does this really seem plausible? I think it is open to two objections:
- I see no reason to think that the notion of mathematical expectation is really a measure of anything in the world. Suppose it is true that there is something called 'logical expectation,' and that it is a function of the probability of an event and of the most probable measure of its goodness, is there the slightest ground for thinking that this function is the product of the two? Is it not merely another case of that unjustifiable simplification which in the Hedonic Calculus assumes:
- that there is such a thing as quantity of pleasure, and
- that it must be measured by the time integral of the intensity of a pleasure?
- Is the definition of objective rightness which we have reached as a matter of fact what anyone means by rightness? I quite agree with Mr. Russell that all ethical phrases are used ambiguously by common sense, and, therefore, whenever we try to give a strict meaning to them, we shall meet with verbal paradoxes. Still, we must not get entirely away from common sense, but try to be clear as to the various separate concepts which it verbally confuses. It is because of certain judgments of common sense that Russell introduced his theory of objective rightness, and it is a real objection to it that it is not only infected with all the doubt and vagueness that attach to the notion of expectation, but also seems hardly to correspond to any of the senses of rightness which common sense vaguely recognizes. If, then, any plausible alternative can be offered, I should be inclined to prefer it.
Let us then leave Russell's theory about objective rightness and consider Moore's. This theory makes objective rightness turn solely on the actual goodness of actual consequences, whether they are probable relative to the agent's information or not. It has then to deal with the apparent paradox that, whilst unforeseen circumstances may cause the actual consequences of an act to be utterly different from what could have been expected, we do not blame a man because what he has done, on the logically justifiable expectation of its having good results, turns out to have bad ones. Moore's answer is that the paradox arises from confusing what is right to do with what is right to praise, or holding that it is only right to praise right actions and only right to blame wrong actions. This supposition is not necessarily true. A's praise or blame of B's act is a second act and, like all others, its rightness or wrongness must be judged by its own consequences and not by those of B's act. It will be right for A to praise B's act if the consequences of doing so are at least as good as those of blaming it or saying nothing, no matter what the consequences of B's act may prove to be.
This theory seems to me very plausible, and I think Moore is right in saying that much of the paradox is merely apparent. It is, therefore, worth while to consider the question of praise and blame more closely. The words praise and blame are somewhat ambiguous, and it is important to distinguish three elements:
This ambiguity leads to an ambiguity in the question: Am I right in praising some acts that are really wrong, and blaming some that are really right? This may mean:
- the judgment that an act is right or wrong,
- certain peculiar feelings of approbation and disapprobation, and
- the expression of such judgments and feelings.
In my opinion, 'ought' has a different meaning in each of these questions. The first means: Am I ever logically justified in holding that an action is probably wrong when it is actually right, or probably right when it is actually wrong? The answer is clearly in the affirmative. I ought logically to believe probable on my information what actually is probable, but this may be the opposite of what is actually true.
- Ought I to believe that some wrong actions are right, and that some right actions are wrong?
- Ought I ever to have the feeling of approbation to a wrong action, and that of disapprobation to a right one? and
- Ought I to assert what I believe, and state what I feel in such cases?
It is perfectly clear that in the second question 'ought' cannot have this meaning. I am under no logical obligation to have any particular kind of feeling in given circumstances. But it is also clear that it cannot simply have the meaning that it is right for me to have such and such a feeling in Moore's sense of right. In the first place, for an act to be right in Moore's sense, it must be voluntary, whilst, whether I have a certain feeling under certain circumstances, is largely independent of my will. Further, rightness for Moore depends on goodness of consequences. Now my unexpressed feelings (using expression in a sense wide enough to include a frown and a philippic) have few consequences outside myself. I admit that they may be important; but what I want to suggest is that there is undoubtedly a sense of rightness in which it may be said that a certain feeling is the right one to have, under certain circumstances, no matter what the consequences may be. Take, for instance, the emotion of sorrow on the death of a friend. It is not a voluntary product, and therefore not right or wrong in Moore's sense. But if it could be right or wrong in this sense, it would almost certainly be wrong, for it is difficult to see what good consequences can come from sorrow at what cannot be altered. Nevertheless, there is a perfectly definite sense in which we should say that sorrow is the right, and hilarity the wrong emotion under the circumstances. And this is a new use of right and wrong. Sorrow is not good, nor are its consequences as a rule good, but it may be right. There is, of course, a connection between this sense of right and wrong and goodness and badness. What is called right or wrong here is a feeling in connection with a situation. I think this sense of right might be defined as follows: The feeling x is right in the situation y, when the complex xy is intrinsically better than x alone, and at least as intrinsically good as the complex formed by y and any other feeling that can be directed toward it. Can is not used here in the sense of 'can if one will.' In certain circumstances I ought to have certain feelings, whether I could have them by willing or not.
Our second question, then, comes to this: Is the feeling of approbation ever the right feeling (in this new sense) to have toward acts that are wrong in Moore's sense, and that of disapprobation the right feeling to have toward one that is right in his sense? Here again the answer is in the affirmative, but needs some qualification. The sort of feelings that are right or wrong are those directed to an object, as, for instance, sorrow is always sorrow for something, and approbation approbation of something. And feeling can only be directed to objects as they are known or believed to be by the person who has the feeling, not to aspects of the objects about which he knows nothing or is misinformed. Thus, the right feeling toward an act may very well alter as time goes on and more is known about its consequences; there is, of course, nothing paradoxical about this, because at each different stage of knowledge and belief the
feeling is really directed to something different. Further, in practice, our feelings are never toward an act alone, but we have a total feeling that depends on two factors:
As far as I can see, the right feeling toward any act that exhibits conscientiousness is pro tanto approbation, though we may believe that the results of the act will be so bad that our total feeling ought to be condemnatory. The following, then, seems to be the answer to our present question: An act which is wrong in Mr. Moore's sense ought to meet with feelings of approbation by anyone who believes that the intrinsic value of the whole, formed by the consequences and the moral qualities which the choice of the act exhibits in the agent, is at least as great as that of the whole formed by any alternative act + the moral qualities that it would have exhibited. It may thus be right for us to feel approbation for an act that is not merely wrong in Moore's sense, but is believed to be wrong by us. The moral quality that seems most to add to the values of such wholes is conscientiousness, so that it is often right to approve an act that is wrong and is believed to be wrong, because it has been conscientiously performed. Of course, it is open to anyone to say that here we really have two different feelings, one directed to what we believe to be the rightness or wrongness of the act, and the other to what we believe to be the moral qualities that it exhibits in the agent. I somewhat doubt personally whether, when two objects are so closely connected as an act and the volition that produces it, you can analyze the total feeling directed toward them into two feelings, one directed to each object; but the point is not of great importance for the present purpose.
- on our belief in its rightness in Moore's sense, and
- on the moral qualities of the agent which we infer from the act.
Finally, we come to the third meaning of 'ought' involved in Moore's theory in the question: Ought I to assert what I believe and state what I feel about the rightness of acts? Here, of course, the meaning of ought corresponds to Moore's general meaning of right, The question is: Will the consequences of stating my belief that an act is right, and showing my approbation of it, have at least as good consequences as any alternative action even when as a matter of fact my belief is false? The answer, of course, is that it probably is sometimes objectively right to praise an objectively wrong act, and vice versa. But if Moore means to offer as a definition of subjectively right acts that they are those which it is objectively right to praise, the definition will hardly do. In the first place, it will clearly follow that the extreme doubt that attaches, on his theory, to whether any particular act is objectively right will now equally infect the question whether it is subjectively right, since the subjective rightness of all acts will depend on the objective rightness of a certain class of acts. Moreover, it is easy to imagine acts which it is almost certainly objectively right to praise, and almost certainly subjectively wrong to do. If my friend and I were in the hands of a cruel despot with a taste for flattery, it might very well be objectively right for me to praise his wicked acts in order to save myself and my friend from his cruelty; but this would not make his wicked acts even subjectively right if he believed he was doing wrong.
Let us, then, go a little further into the question of praise and blame for ourselves, since neither Moore nor Russell have descended to detail. Before we go any further it will be well to compare Moore's and Russel's views about probable goodness for a moment; for it is becoming evident that Moore's view is going to introduce probability as well as Russell's, though in a different way. The difference is this: On Russell's view, a probably most fortunate act is an actually right one; on Moore's view, a probably most fortunate act is a probably right one. And this, being quite general, applies, of course, to acts of praise and blame. Now, I think there is some risk of the inconsistency of using Russell's definition of rightness for acts of praise and blame, and
Moore's for other acts, and defining subjective rightness by a mixture of the two. This inconsistency must, of course, be avoided if we want to find the real consequences of both views. The consequences of applying Russell's definition are somewhat complicated. We will consider them first.
It might seem at first sight that, on Russell's definition, it will always be right for me to praise what it is right for me to do. For to say that the act X is right for A to do is to say that it is a probably most fortunate act on his information. Now, praise of such acts will certainly tend to make other people perform them, and it seems as if it must be right for A to make other people cause results which it would be right for him to cause himself. But this does not by any means follow. X may be that action which, on A's information, is probably the most fortunate of any that are open to him, but it does not follow that it is probably the most fortunate of any that are open to B; for B may have alternatives open to him which A has not. Hence A's praise of what it would be right for himself to do may cause B to choose an alternative which, on A's information, is not the probably most fortunate one open to B. There is, too, another consequence that is worth mentioning. Suppose that B has only the same alternatives open to him as A, but different information from A; then, on B's information, X may not be the probably most fortunate act open to him, though it is on A's information. Hence A's praise of what it would be right for himself to do may cause B to do what is wrong for him to do.
Let us now apply Moore's theory of rightness to the question of praise and blame. I do not see how we can dispense with the notion that there is a definite sense in which it is right to do what we believe to be objectively right, by means of any considerations about praise and blame. I take it that Moore is trying to make us believe that the notion of subjective rightness can be resolved into that of objective praiseworthiness. This might
mean one of two things. It might mean that the two notions are really identical, that the second is an analysis of what people mean by the first; or that, whilst they are genuinely different notions, they have precisely the same extension. The first alternative, of course, implies the latter part of the second. It seems to me that inspection shows that the first alternative is false; if we can further prove that the second is wrong, we shall have an additional refutation of the first. It seems to me that when I have done my best to determine which alternatives open to me are objectively right, I can well admit that I may be mistaken, and yet be certain that it is right for me to do one of these alternatives. And I certainly do not seem to mean by this that it will always be objectively right for me to praise myself or for other people to praise me. For one thing, I do not generally think about praise or blame at all when I think about rightness; and for another, I can no more be certain about the effects of praise and blame than about those of any other action. Further, the sense of rightness which we are trying to analyze is, I think, essentially connected with conscientiousness. Now, I grant that on Moore's theory it is probably objectively right for me to praise what it is probably objectively right for me to do, because I shall thus tend to cause actions that are probably right. But the probable rightness of my praise is independent of the motives of the actions praised, since it merely depends on the probable goodness of their consequences, and this is unconnected with their motives. Thus, the probable rightness of praising an action has no immediate connection with its motive. On the other hand, whilst I should call an action subjectively right that was done conscientiously, it is not clear that it would ever be probably right for me to praise it on this ground. By praising it I might promote conscientiousness; but if it is probably an objectively wrong action, I should also probably be praising objectively wrong acts. I could avoid this difficulty, of course, by praising the motive and blaming
the action; but it is not at all clear to me that, on Mr. Moore's theory, it is ever probably right to praise conscientiousness. If it be probably right to praise conscientiousness, it must be because that quality is either good in itself, or adds to the goodness of other wholes, or is likely to have good results. Now, I understand Moore to hold that motives have no intrinsic value, which cuts out the first alternative. As far as I know, he has not expressed any opinion about the second alternative. Finally, I do not see the least reason to suppose that conscientiousness is more likely to produce objectively right acts, on his theory, than any other motive. No doubt, it involves that you have done your best to find out what is right and are going to act on your conclusions; but, since the rightness of your action is at the mercy of all that is going to happen in the universe throughout all future time, there is no reason to expect better results from conscientious acts than from the most stupid and biased ones. I conclude, then, that, since the notion of subjective rightness is essentially bound up with conscientiousness, whilst the rightness of praise and blame is directly connected with consequences and not with motives, and further since there is no ground for supposing that it is ever probably right to praise conscientiousness rather than other motives, the notions of subjective rightness and of objective praiseworthiness cannot coincide either in meaning or in extension.
It remains for me, after these criticisms, to see whether anything positive can be said. I agree with Russell that it is neither important nor possible that the terms which we use should always exactly coincide in extension with those used by common sense. All that is important is to recognize clearly every different notion that is involved in the judgments of common sense; to give separate names to them and use them consistently; and, where possible, to analyze them and determine their mutual relations. It may also be necessary for us to recognize notions which common sense does not, and to determine
their relations to those which it does. Let us then consider the notions with which we have to deal.
In the first place, Moore's notion of rightness, which is the same as Russell's 'most fortunateness,' is a perfectly definite concept which anyone is at liberty to use. My only objection to it is that it involves a somewhat arbitrary cut out of a wider notion, a cut which, I think, is made at a different point from any at which common sense makes one. What I mean by the wider notion is the intrinsic value of the whole universe, past, present, and future. Moore and Russell cut this in two at a voluntary decision between alternatives, and consider that rightness is only concerned with the states of the universe, after this decision. They then further cut the total state of the universe, after the decision, into the part that is and the part that is not a consequence of the decision, and they connect rightness only with the former of these two fractions. We are not told precisely how this second decision is made; but I think the consequences of an act are taken to be everything in the universe that would not be the same, whether the act had been done or not. There is some ambiguity in this, however. Do you mean the same in fact or in value ? If you mean in value, you cannot safely take in less than the total future state of the universe as the consequence of any act, because its value will be different according as the act is done or not, provided that the doing or withholding of the act makes a difference to any part of it. This is evidently not what is meant. What is involved is only those facts that would have been different as facts if the action had not been done, and the value of the whole thus formed.
From each of these two ways of cutting up the universe there follows a result that has not, I think, been noticed. It is that a most fortunate act may make the total state of the universe worse than a less fortunate one. If I do x, I may make the total future state of the universe better than if I do y; but the Principle of Organic Unities
precludes us from asserting that, because the state of the universe, after a moment t, is intrinsically better if I do x than if I do y, therefore its total state, before and after t, is better. For, if two wholes consist of a common part and two different parts, it does not follow that that which has the better part is as good as that which has the worse one. So much for the results of the cut in time. But it follows for similar reasons that the cut in the future states of the universe makes it possible that the total future state itself may be worse through a right act than through a wrong one. The consequences of x may be better than those of y; but the whole made by the consequences of x and the rest of the future states of the universe may be worse than the whole made by the consequences of y and the other future states of the universe. These possibilities will equally arise wherever you make the cut. Since it seems paradoxical to say that a most fortunate act may make the total state of the universe worse than a less fortunate one, I shall define a most fortunate act as one such that, if it be performed, the total state of the universe will be at least as good as if any other act open to the agent were performed. It follows that motives will be relevant to the question whether an act is most fortunate, for they are parts of the universe that precede the act and make their contribution to the total value. I think, further, that it has considerably more claim to be called a right act than what Moore calls a right act. By this I simply mean that, whilst it shares with Moore's meaning of right the objection (if it be an objection) that rightness and wrongness then depend on unforeseeable circumstances, yet it is closer to one part of what common sense means by rightness than Moore's definition. Suppose we put aside the question of unforeseeable circumstances by considering the acts of an omniscient God, a conception familiar enough to common sense. Common sense would say that all God's acts are right and that they all produce the greatest possible good on the whole, and it would hold that the two statements are necessarily connected. But with Moore's definition we have seen that there is no necessary connection between them, and the question could arise whether God ought to do what is right or what makes the total state of the universe as good as possible. Since it seems clear that he ought to do both, it looks as if the two notions must coincide in the case of an omniscient God at any rate, i. e., when we leave out of account the question of intellectual limitations. I suggest, then, that the places where Moore and Russell make cuts in the total state of the universe are really arbitrary and do not correspond to any distinction involved in the judgments of common sense, nor, so far as I can see, to any that is of ethical importance.
I shall, therefore, define an objectively right action as one such that, if it be done, the total value of the universe will be at least as great as if any other possible alternative had been done by the agent. I have now to consider how far this agrees with, and how far it differs from, the meaning of the word involved in the judgments of common sense. It agrees, as we have seen, entirely when intellectual limitations are set aside. It also allows of agreement in the matter of motives. Common sense attaches a very great weight to motives, though not, I think, an exclusive one. This is quite in accordance with our theory. The fact that an agent does a certain act from a certain motive may be so valuable as to outweigh the badness of its consequences in Moore's and Russell's more restricted sense of that word. The total state of the universe may be much better if I do an action which will have very bad effects from a sense of duty, than if I do an alternative which will have much better effects from a desire to give pain. On the other hand, it is always possible to imagine consequences so bad that no goodness of motive will balance them. This seems to me in complete accord with common sense. Torquemada's actions were almost certainly wrong, in spite of the goodness of his motives, because of the extreme badness of their effects in Moored and Russel's sense; Pitt's action in financing the early stages of the revolutionary war by loans may have been right because of the goodness of his motives, in spite of the considerable badness of the results.
Thus, motives actually enter into objective rightness on our definition, as I think they do for common sense. But I do not maintain that this sense of rightness completely agrees with all usages of the term by common sense, or that it furnishes a complete account of the common-sense attitude toward motives. There may be marginal cases where common sense judges an action to be right even though it thinks it probable that, when both motive and consequences are taken into consideration, the total state of the universe is worse than if the agent had chosen another alternative. In such cases, however, I think it would tend to say, not that the action was right, but that it was right of the agent to do the action. Thus, common sense might well say that Torquemada's actions were wrong in spite of the goodness of his motives, but that it was right for Torquemada to do them. It is not consistent in its use of the terms, but I think it is quite clear what is meant. To say that Torquemada *s actions were wrong does not merely mean that the consequences apart from the motive were bad, but that the bad consequences plus the good motive were bad. To say that Torquemada was right to do them is to pass a judgment on Torquemada's motives in abstraction from the total results of his act. As a matter of fact, the two judgments are likely to be made together; they much more often agree than conflict; and they both involve motives, the one partially and the other exclusively. It is not, therefore, surprising that common sense is liable to confuse them.
Common sense calls an action right for the person who does it, when it approves of the motive; and it will be well worth while to consider motives for a moment. In the sense in which I am using the term, we cannot say with Russell that a motive is simply the cause of a decision. Probably all decisions have causes, but they have not all motives. When I say that I do an action because I judge it to be right, I do no doubt imply that my belief is a part of the cause of the decision; but I am not clear that this is all that I mean by the phrase 'because of' here. Anyhow, motives are a special class of causes of decisions of which the following things can be said: (1) They involve beliefs in the qualities or effects of the act which they cause to be chosen, and (2) the belief has to be quite explicit, and has to be explicitly recognized as in some sense the last and completing factor in the cause of the decision. Now, there is one kind of belief to which common sense attaches very great ethical importance as a motive, viz., the belief in the objective rightness of the action. Common sense considers it a supreme excellence of character when beliefs in the rightness of acts are habitually causes of deciding to do the acts. And it marks its approval by saying that it is always right for a man to act from this motive even though the action be wrong. The excellence of this motive will indeed often make actions done from it objectively right, in spite of the badness of their effects; but even when the effects are too bad for the goodness of the motive to counterbalance them, common sense will still say that it was right for the man to have acted as he did. And this sense of rightness is peculiarly connected with this kind of motive. Thus, common sense, as distinct from Kant, recognizes that some actions are better when they spring from other motives or from no motive at all; it is better, for instance, other things being equal, to be kind to people because you like to see them happy than because you judge that it is right to make them happy; but, if the act is judged to be objectively wrong in a particular case, it will be said to have been right for the man who did it because he thought it right, and not for the man who did it from a direct desire to give happiness. We rightly prefer the action of a man who spoils his children because he likes to see them happy to that of one who spoils them because he is a conscientious hedonist; but we should say that the action was right for the second and wrong for the first.
It is clear that this sense of rightness corresponds closely with Russell's subjective rightness. But there is one point where I think Russell makes subjective rightness too subjective. He says that it is subjectively right to do what is conscientiously believed to be objectively right, but that does not imply that to be subjectively right a man must hold Russell's theory as to what is meant by objective rightness. This, I think, is a mistake due to natural modesty. If Russell's theory of objective rightness be true, then a man is not subjectively right if he means something else by objective rightness, and does what he judges to be objectively right on his own definition. In fact, a man is not subjectively right unless he holds true views as to the meaning of objective rightness. He may be as mistaken as he likes as to whether an act really has the necessary qualities for making it objectively right, but unless there be agreement as to what these qualities are, there is nothing in common to those beliefs, by agreeing in which men are called subjectively right. You cannot avoid this by saying that it is subjectively right to do that for which you have a feeling of approbation. When feeling of approbation and judgment of objective rightness go together, this is true; but, when they diverge, it is not even subjectively right to act for the feeling and against the judgment. Thus, it is always the judgment that is relevant to subjective rightness; and, therefore, if there is to be a definite common meaning for subjective rightness, there must be a definite common meaning for objective rightness. Thus, Russell ought only to call a man subjectively right in his conscientious actions if the man attaches the same meaning as he himself does to objective rightness.
The upshot of the discussion, then, is as follows: (1) I consider Moore's definition of objective rightness and Russell's definition of a most fortunate act too narrow. They make an arbitrary cut in the whole universe and
lead to paradoxes which make us think that they cannot be a part of what common sense means by rightness. (2) I do not think that subjective rightness has any very close connection with the objective rightness of praise or blame, but that it is particularly connected with those motives which are called conscientious, taken in abstraction from the rest of the act. (3) But Russell's definition of subjective rightness is too subjective; for any agreement about subjective rightness involves an agreement about the meaning of objective rightness. (4) There is a special sense of rightness which applies to feelings as directed to situations. (5) I doubt whether common sense means by objective rightness what Russell does, and I hold that his account remains obscure, partly because you cannot talk of the probability of a proposition, and partly because of the ambiguity of the phrase 'a probably most fortunate act.' (6) My definition of objective rightness agrees with common sense in making motives an actual and important factor in it; and I think that the difficulty about the attitude of common sense to wrong acts done through ignorance of unforeseeable circumstances is met by the view that it is subjectively right for the agent to do what he judges to be most probably objectively right on his information. (7) But finally the existence of three logical senses of right and ought does make it just possible that there is an ethical sense of right corresponding to Russell's objective rightness, though the difficulty as to what precisely is meant by a probably most fortunate act infects this possible meaning with ambiguity.
1 Russell says 'the best consequences,' and speaks of 'the most fortunate act'; but Moore rightly points out that two possible acts may have equally good consequences, and that they will then both be right if the consequences are better than those of any possible alternative. I have altered Russell's language to meet this objection.