IMPERATIVES, INTENTIONS, AND THE LOGIC OF "OUGHT" (original) (raw)

A revised and enlarged version of "Imperatives, Intentions, and the Logic of 'Ought'," Methodos 8 (1956), in Morality and the Language of Conduct, edited by Hector-Neri Castañeda and George Nakhnikian (Wayne State University Press, 1963): 159-214.

IMPERATIVES, INTENTIONS, AND THE LOGIC OF "OUGHT"
Wilfrid Sellars Yale University

INTRODUCTION

My purpose in this paper1 is to explore the logic of "ought," with a view to determining the relations which obtain between it and certain other key terms of practical discourse. In particular, I shall be concerned with the relation between "I ought to do X" and "I shall do X"; "You ought to do X" and "Do X!"; "He ought to do X" and "He shall do X"; and between "I ought to do X" and "He ought to do X." I shall also be concerned with the relation between "Everybody ought to do A, if in C" and "Would that everybody were happy!" on the one hand, and "Would that I were happy!" on the other.

Put in more traditional terms, this paper is about the relation between "thinkmg oneself under an obligation to do something" and " (ceteris paribus) deciding to do it for the reason that one ought to do it," and between "being under an obligation to do something" and "having a reason for doing it, namely, the reason that one ought to do it." Insofar as I have a thesis to defend it is the twofold one that to know that there are certain things that one ought to do is to have a sense of duty, and that obligation, by its very nature, is intersubjective. And if, thus baldly stated, neither component of this thesis is new (for the first smacks of emotivism and second of intuitionism), some interest may attach to the way in which this joint thesis is developed, and the consequent clarification of the relation between the sense of duty on the one hand, and the impartial love of humanity on the other.

If one were to ask a convinced student of The Right and the Good.2

Could a person ''apprehend a principle of prima-facie obligation" to the effect that one ought, for example, to keep promises -- that is, Anglice, could one know that, other things being equal, one ought to keep ones promises -- and yet have no tendency to be moved to act on specific occasions by the thought that one ought to do this as being the keeping of a promise?

-- would he not answer (though I must qualify this in a moment) that there is no contradiction in the idea of such a person, or (if he should think that tending to be so moved is a defining trait of a person3) at least in the idea of such an intelligent animal? But if he finds no contradiction in this idea, would he be content to say that it is a contingent empirical fact about human nature that people who know what they ought to do tend to be moved to act by the thought that they ought so to act? Or would he not be more likely to subsume this fact under the heading of the "synthetic a priori"?

It must be admitted, and this is the qualification mentioned above, that some intuitionists. notably H. A. Prichard. seem to think that it is not a merely empirical mistake but, rather, an absurdity to suppose that one might apprehend what he ought to do, yet have no tendency whatever to be moved to act by the thought that he ought to do it. Thus, when Prichard writes, "To feel that I ought to pay my bills is to be moved towards paying them."4 the context makes it clear that he is not thinking of the feeling that I ought to pay my bills as something distinguishable from (but built upon) the apprehension that I ought to pay them, but rather as being this apprehension itself. In comparing the apprehension of obligation to the apprehension of mathematical truths, Prichard writes that "in both [cases] insight into the nature of the subject leads me to recognize its possession of the predicate."5 but he nevertheless insists that the apprehension of an obligation is different from the apprehension of a mathematical truth. My difficulty with Prichard on this point (as with the many philosophers who have insisted that moral thinking qua thinking is "conative," no "mere blend" of thinking and "emotion") is that he offers not even the beginning of a satisfactory analysis of this phenomenological insight, an analysis which would account for the fact that moral thinking differs from, but resembles, other forms of thinking, by relating both to the fundamental categories of an adequate philosophy of mind. Prichard's grasp of the distinctive traits of moral thinking exhibits the combination of thinness and acuteness which is characteristic of his philosophy. It is because Sir William David Ross has turned thin truth into quick error by forcing Prichard's insights into the procrustean bed of neo-Aristotelian theory, that we could picture, as we did. the response of the student of The Right and the Good to our question about ought and motivation.

If, now, we turn to a convinced student of Language, Truth and Logic6 and ask.

Can a person know that he ought to do this action without being moved to do it?

I think it reasonably clear that his answer would take the form of a commentary on the question. The upshot would be some such statement as the following: "The so-called thought that one ought to do A here and now is not, strictly speaking, a thought at all, but rather a specific way of being moved to do A. With this qualification the answer to your question is no, for the simple reason that to 'know' that one ought to do A here and now is to be moved to do it." If we press him to say just how the specifically moral character of the "being moved" is to be understood, if it is not a matter of being moved by the thought that one ought to do A, it is unlikely that we would get a satisfactory answer. For in its early stages emotivism, naive or sophisticated, views as analytic the connection between "thinking that one ought" and "having a motive" which classical intuitionism, with the exception we have noted, takes to be either empirical or synthetic a priori. On the other hand, as is well known, it tends to view thinking that one ought as "thinking" that one ought.

It was the signal merit of intuitionism, particularly of the deontological variety, to have insisted on the uniqueness of prescriptive discourse, as over and against traditional naturalism's attempts at reduction, and on the truly propositional character of prescriptive statements, as over and against the emotivist contention that ethical concepts are "pseudo-concepts" and the logic of moral discourse a "pseudo-logic." But the epistemological and metaphysical commitments of ethical intuitionism, which precluded it from understanding the logical connection between "thinking that one ought" and "being moved to do," thus forced it to make a mystery both of the conduct-guiding role of moral discourse, and of the uniqueness of prescriptive discourse which it had so happily emphasized.

The situation seems clearly to call for a theory which, without denying that ought-statements stand, as such, in logical relation to one another, makes the connection between moral thinking and doing analytic, a matter of strict logic. That attempts along these general lines are in the air is clear. One of these, the neo-imperativist approach of R. M. Hare,7 is, in my opinion, sufficiently close to the truth to be a useful point of departure for the ideas I wish to develop. While I think that something like his account of the concept of ought is true, I do not think that it will do as it stands. Indeed, I think that its very closeness to the truth has enabled it to obscure essential points about the concept of obligation.8

Universality and the Logic of Imperatives
1

A person who says, "X ought (morally) to do this," commits himself to supporting this statement with a statement of the form, "Doing this, in these circumstances, is doing A in circumstances C, and anyone ought always (ceteris paribus) to do A in C"9 This fact is sometimes put by saying that singular ought-statements imply universal ought-statements. This is true, however, only in a special sense of "imply." Nor will it quite do to speak of "presupposing," for though one who makes a singular ought-statement is committed to support it with a universal ought-statement, it is not necessary that he have the latter "up his sleeve." If a wise man tells Jones that he ought to do a certain action, he has a reason to think that he ought to do it, and that the action could be subsumed under a moral principle, even though he is not in a position to do so. It is for this and related reasons that I have elsewhere10 characterized this sense of implication as dialectical. Yet, though it is a "looser" relation than presupposing, let alone ordinary implication, it is, like the latter, a logical, as opposed to psychological, relation.11

Now even if there is no plausibility at all to the suggestion that "You ought to do B" is simply the imperative "Do B!" with or without psychological embellishments, the claim that the ought-statement is equivalent to a singular imperative which is implicitly universal in the above sense is something to be reckoned with. This claim can be construed as the idea that "You ought to do B" is equivalent to "Do B!*" where the asterisk is a signal of a dialectical commitment to support it with an "argument" of the form,

Let everyone universally do A in C!
Doing B is doing A in C
So, do B!

Thus, the claim is that in "You ought to do that," the ought plays the dual role of (1) giving the infinitive ("to do") which follows it the force of a verb in the imperative mood ("Do!"), and (2) embodying a commitment to support this singular imperative in the manner indicated above.

Now this claim (which is, of course, only a rough approximation to more sophisticated analyses of the imperativist type) is open to many objections, some of which, in my opinion, are sufficiently compelling to call, at the very least, for a substantial revision. It will be useful to begin our exploration of the logic of ought by raising one of these objections, the full force of which will not emerge until a later stage of our argument. This objection is that the "ought" in the universal ought-statement "Everybody ought always to do A in C" does not seem to be redundant. Yet it cannot be accounted for in terms of the two roles of ought specified above. For the latter, (2), that of signaling a dialectical commitment to back up the sentence in which it appears with a sentence having the logical form of universality, is on the face of it irrelevant, in the case of the first principles of obligation, while the purpose of the former, (1), can easily be achieved by using the imperative mood to start with, that is, by simply saying, "Let everyone always do A in C!"

The answer of one imperativist to this objection is to be found in the concluding pages of The Language of Morals. There Hare argues that in addition to playing roles which are essentially the same as those indicated above, the ought also gives the imperative it signalizes a truly universal force. He points out that in actual usage, "Let everybody always do A in C!" means "Let everybody from now on do A in C!" (it being silly to say, for example, "Let everybody wear blue suits yesterday!"). "Everybody ought always to do A in C" has, as Hare sees it, an imperative force which is suggested by the absurdity of "Let everyone do A in C throughout the past as well as the future!" in actual usage.

2

To evaluate Hare's reply, we must begin by considering certain features of the logic of commands.

The issuing of commands (i.e., commands only, not sentences in the imperative mood generally) is not a promiscuous activity. Not everyone on every occasion can properly command another to do a certain action. There must be something about one person's relationship with another which authorizes him to issue the command; and a relationship which authorizes one person to command another to do something may not authorize him to command something else. I referred above to the issuing of commands as a performance, and this is, of course, the heart of the matter. The parallel with promising is instructive. As Austin has pointed out,12 making a promise is a performance which creates a presumptive prima-facie obligation to do A on the part of the person who says, "I promise to do A." It "creates" this obligation by virtue of the moral principle which can, for our purposes, be formulated as follows:

If X appropriately says "I promise to do A" to Y, then X owes it to Y to do A.

Now, whereas promising is a performance which binds the speaker, issuing a command binds the person to whom it is issued. Thus, issuing a command within the limits of one's authority "creates" a presumptive prima-facie obligation to do the action commanded on the part of the person to whom it is addressed. And this performance "creates" an obligation, binds, only because, like promising, it rests on a principle, in this case:

If X appropriately says to Y "I command you to do A! then Y owes it to X to do A.

Thus the claim which commands have on our obedience is but a special case of the claim which our obligations have upon us. Obeying a command, like keeping a promise, is a special case of doing one's duty -- though to characterize any particular obeying or promise-keeping as a doing of one's duty is, of course, a defeasible rnatter.13

There is another, if closely related, respect in which the logic of commands resembles the logic of promises, and it is this respect which is directly relevant to our argument. Consider the sentence:

I promise to call you, if it rains.

There is no such reasoning as

I promise to call you tomorrow, if it rains tomorrow
It will rain tomorrow
So, I promise to call you tomorrow

or, in general,

I promise to do A in dt, if p
p
So, I promise to do A in dt.

It is, for reasons which will be developed in a moment, nonsense to suppose that one performance of promising can, as performance, be a conclusion from another. A person who believes that p and intends to do A, if p, may promise to do A, or, more cautiously, to do A, if p. His reasoning in foro interno as far as his intentions are concerned may perhaps be represented by the schema:

I shall do A in dt, if p
p
So, I shall do A in dt.

But while there would be an absurdity in saying "p and I shall do A, if p but I shall not do A," there seems to be no such absurdity in the saying "p; and I promise to do A, if p; but I do not promise to do A." (KB: "do not promise" must not be confused with "have not just promised.") It would, however, for reasons pertaining to the relation between promising and intending, be absurd to say "p; and I promise to do A, if p; but I promise not to do A."

3

The opening remarks of the preceding section make it clear (if it was not already clear) that there would be a logical howler in any attemgt to "reduce" ought-statements to commands. It might seem equally clear that to show this is not to show that ought-statements cannot be "reduced" to imperatives. For although commands, like promises, presuppose principles of obligation, surely, it will be said, simple imperatives do not. Telling someone to do something does not as such appear to create an obligation on his part to do it. On the other hand, the fact that you told him to do it (as contrasted with the possibility that he might merely have overheard you "intend out loud" that he do it, or even that you might have told him of your intention that he do it) is relevant in a special way to his deliberations on his rights and duties, particularly vis-a-vis you, and on whether he has good reasons for doing what he proposes to do. Intending out loud, telling of your intentions, and telling someone to do something are all of them pieces of conduct. They must, however, be carefully distinguished from one another, for they enter in different ways into the web of prescrptions and norms which govern our relations with our fellow men. Indeed, while it may be literally true that the fact that you tell someone to do something does not create an obligation on his part to do it. It may well presuppose a context of obligations for its significance as telling. While simply telling someone of your intention to do something does not create an obligation on your part to do it -- at least in that simple way which characterizes the institution of The Promise, nevertheless it would be misleading to say that telling one's intentions to somebody is logically prior to or independent of practical principles. Again, if X stands in certain relations to Y, the fact that X tells Y to do A generates a claim on Y to do A. It can even be argued that in the case of the highwayman who brandishing a pistol, says, "Give me your money!" the ought of prudence is being mobilized, and that, in general there are no proper imperatives without some connection with practical principles, or, at least, that in such cases telling degenerates into merely making manifest one's desires and intentions.

I have contrasted "intending out loud" that X do A with telling someone (perhaps X himself) of one's intention that X do A. Both of these seem to be covered by the ambiguous phrase "expressing one's intention," though I think it is properly limited to the latter, or "telling of," case. A similar danger of confusion is to be found in the case of the "expression of belief." Here too we must draw a threefold distinction between "thinking out loud that p" (thinking out loud being the basic form of all thought), ''telling someone of one's belief that p" (i.e.. expressing one's belief that p), and "telling someone that p" For our purposes, the crucial distinction in each case is that between telling of (or expressing) one's intention or belief on the one hand, and telling to or telling that on the other. Telling of one's beliefs or intentions, as we are using this phrase, is not to be confused with describing oneself as having these intentions. It should not be overlooked, however, that the self-description "I intend that X do A" (though not its cognates in other persons, tenses, and moods) has the force of an expression of intention in addition to its descriptive role, just as "I think that p" has the force of an expression of "the thought that p" in addition to its autobiographical role.

Before continuing with the argument, there are two terminological conventions I should like to adopt in this paper. The first, a simple matter of convenience, is that whether the subject of a sentence of the form

X shall do A

is "I" or "you" or "Jones," it always expresses the speaker's intention or resolve that X do A. Although this stipulation does minor violence to correct English usage, the awkwardness is more than offset by the gain in simplicity of formulation. Correspondingly, sentences of the form

X will do A

even in the first person are to be understood as expressions of the speaker's belief that the person referred to by "X" will do A.

My second stipulation is that an utterance which not only expresses the speaker's intention that the person referred to by "X" do A, but plays the telling to role as well, is to be represented by the form

X shall do A!

where the "!" signalizes this role. Thus, in terms of this convention, the utterance

X shall do A

simply expresses the speaker's intention that the person referred to by "X" do A. It may be overheard by this person, but it is not, as overheard, a telling him to do A. It might appear that the form

You shall do A (as contrasted with "You shall do A!")

must be ruled out on the ground that utterances in the second person are not overheard by the person addressed as "you." That this is not the case will become clear from what follows.

The concept of telling someone to do something is to be distinguished from that of telling someone one's intention that someone (perhaps the same person) do something. Thus, to say "I shall do A" to X is to tell X of one's intention to do A, but is not to tell anybody, even oneself, to do A. Correspondingly, to say to X "You shall do A," while it is at least a telling X of one's intention that he do A, need not have the force of "You shall do A!" i.e., the force of telling X to do A. Again, while

Tom shall do A

simply expresses the speaker's intention that Tom do A,

Tom shall do A!

said to Tom tells him to do A, and said to Dick, will, according to the conventions of this paper, have the force of the imperative, "Let Tom do A!" which not only tells Dick of one's intention that Tom do A, but tells Dick to do something, roughly what he reasonably can to ensure that Tom does A, and is, therefore, the equivalent, in terms of our convention, of

You (Dick) shall do what you reasonably can to ensure that Tom does A!

Furthermore, imperatives of the form

Let it be the case that p!

will be represented in terms of our convention as

It shall be the case that p!

which, as telling to, would differ from both

It shall be the case that p

as addressed to certain persons and advising them of one's intention that it be the case that p, and

It shall be the case that p

as a simple expression of one's intention that it be the case that p.

A parallel convention will enable us to distinguish

S is P!

as telling someone that S is P, from "S is P" as a telling of (or expressing of) one's belief that S is P, and, a fortiori, from an utterance ot "S is P" as a mere thinking out loud that S is P.

We are now in a position to clarify our intuitive feeling that there is no such reasoning as

I promise to do A in dt, if p
p
So, I promise to do A in dt.

This clarification will also make manifest that there is no such reasoning as

Do A in dt, if p!
p
So, do A in dt!

i.e., in terms of our convention,

You shall do A in dt, if p!
p
So, you shall do A in dt!

Also that there is no such reasoning (in terms of our "telling that" convention) as

q if p!
p!
So, q!

Though there are, of course, the reasonings

X shall do A in dt, if p
p
So, X shall do A in dt

and

q if p
p
So, q.

The point is a simple one about the concept of reasoning. Promising, telling to, telling that, and telling of are all public performances which require an audience. Reasoning is something which can go on in foro interno; and if it goes on out loud, it is the sort of thing which is overheard. It can also be expressed in a sense which parallels the expressing or telling of one's beliefs and intentions.

Is there anything which stands to reasoning and to the expression of reasoning as "telling that" stands to "thinking that" and to "expressing the thought that," and as "telling to" stands to "intending" and to "expressing one's intention"? A plausible candidate is "arguing." But before we ask whether there are imperative arguings, let us note that not only are there no imperative reasonings in foro interno (which is obvious), there are also no expressions of reasonings which have imperatives as premise or conclusion. Only a confusion between a "telling to" and a "telling of" (or expressing) one's intention that X do A could lead to the contrary conviction. (The current practice of speaking of imperatives and resolutives as though they were coordinate from a logical point of view or even related as genus to species -- "to resolve" being to tell oneself to do something -- involves this confusion). Ir makes sense to suppose that an expressed reasoning could have occurred without being expressed; and if so, there cannot be such things as expressed reasonings the premises or conclusions of which are promisings, tellings to, or tellings that. In particular, there is no such thing as imperative inference.

Is there, perhaps, such a thing as imperative argument? And argument is a performance in which a conclusion is defended by offering reasons which give it logical support or purport to do so. Is there, perhaps, such an argument as:

You shall do A!
because you shall do A and B!

Is there, perhaps, an argument of the form

I promise to do A
because I promise to do A and B

There are, of course, the arguments

I told you to do A
because I told you to do A and B

and

I promised to do A
because I promised to do A and B

but, of course, in the latter there is no telling to or promising. Again, there are the supported performances

I promise to do A
because you have been so kind

and

You shall do A!
because you have been naughty

but these are obviously not arguments of which the promising and the telling to are the conclusions. Rather they point to practical reasonings which have as their conclusions

I shall promise to do A

and

I shall tell X to do A

It is tempting to suppose that since an argument is a performance analogous in certain respects to a telling that or a telling to, its conclusion can be a telling that or a telling to. That this is not the case becomes clear when we reflect that what is defended by the argument

q
because p and (p implies q)

is not the telling someone that q (for while there is a sense in which logic provides a reason for telling people that q when we have told them that p and that [p implies q], it does so via practical principles about "telling that" which compete with other practical principles about "telling that"), but rather the thought (expressed or unexpressed) that q. And once we see that not even the performance of telling someone that q is defended by the argument

q
because p and (p implies q)

so that there is no such thing as the argument

q!
because p and (p implies q)!

it becomes equally clear that there simply are no such things as the arguments

I promise to do A
because I promise to do A and B

and

You shall do A!
because you shall do A and B!

Let me try to make my point in terms of Hare's example, which runs,

Take all the boxes to the station!
This is one of the boxes.
So, take this box to the station!

I do not wish to deny that, as Hare14 has stressed, these three sentences are so related that a person who candidly said to Jones

Take all the boxes to the station! By the way, this is one of the boxes.

and yet, though insisting that he had not changed his mind, refused to tell Jones to take this box to the station, would have convincingly shown that he did not understand either one or the other of these sentences. (This point is not unrelated to one made above about the Iogical absurdity of "p and I promise to do A, if p but I promise not to do A.")

4

We have been arguing that the "telling to" signal, "!," like "I promise," does not belong within the context

---, so . . .

and that the only tellings which are appropriate to this context are tellings of. The fact that it does not make sense to speak of imperative inference ("telling to" inference) has been almost as potent a factor in leading people to suppose that there cannot really be such a thing as practical inference, as the fact that expressions of intention are neither true nor false.

Let me begin a more detailed examination of the logic of practical reasoning by proposing the following analysis of the example of the boxes. The analysis breaks the reasoning down into the following moments:

M(1)

This is one of the boxes
So, (Jones will shortly take all the boxes to the station) implies (Jones will shortly take this box to the station)

M(2)

So, (Shall [Jones's shortly taking all the boxes to the station]) implies (Shall [Jones's shortly taking this box to the station])

M(3)

Shall [Jones's shortly taking all the boxes to the station]
So, Shall [Jones's shortly taking this box to the station]

These moments are to be explained as follows. M(i), when made more explicit, turns out to be the second moment of the sequence:

M(1-1)

(Jones will shortly take all the boxes to the station and This is one of the boxes) implies (Jones will shortly take this box to the station)
So, (This is one of the boxes) implies ((Jones will shortly take all the boxes to the station] implies [Jones will shortly take this box to the station])

M(1-2)

This is one of the boxes
So, (Jones will shortly take all the boxes to the station) implies (Jones will shortly take this box to the station)

Two remarks are necessary: (1) "Implies" means that relation between propositions which authorizes inference.15 It will be assumed without argument that implication in this sense includes physical or natural implication as well as logical implication in the narrower sense. Nothing in this paper hinges on the treatment of inductive generalizations as principles of inference. The reader may, if he prefers, press the analysis to a "deeper" level in which laws are introduced as premises. (2) A distinction must be drawn between "independent" and "dependent" implication. Thus, in the conclusion of M(1-1), the first occurrence of "implies" is as independent implication, while the second occurrence is as dependent implication. A dependent implication is one which presupposes a state of affairs to obtain which is not explicitly asserted by the implication statement itself.

P dependently implies Q

presupposes that there is an R such that

P and R independently imply Q

Thus M(1-1), by a correctly interpreted use of the principle of exportation, as it applies to implication proper (as contrasted with "material" implication), takes us from one logical truth to another, where the second logical truth has as its consequent the dependent implication which, by virtue of the next moment to be considered, becomes the principle in accordance with which M(3) proceeds.

M(2) has the form

(P) implies (Q)
So, (Shall [P]) implies (Shall [Q])

a move which holds where P and Q have a content appropriate to practical reasoning. Leaving aside for the moment considerations of tense and the question whether P and Q admit of analytic or self-contradictory values in the context "Shall [--]," we can lay down as a formation rule for shall-statements that one moves from an indicative statement to a shall-statement by turning the indicative statement into a gerund and prefixing it with the shall-operator. Thus, for example, we would go from

Tom will shortly cross the road

to

Shall [Tom's shortly crossing the road].

Notice that the latter does not imply

Tom will shortly cross the road.

We can, in effect, note a parallel in this respect between the shall-statement and the modal statement

Possible [Tom's shortly crossing the road].

On the other hand, it seems proper to stipulate that

Shall [my shortly crossing the road]

implies

I will shortly cross the road

i.e., that it contains the prediction, just as

Necessary [Tom's shortly crossing the road]

contains

Tom will shortly cross the road.

It is important to note that according to the proposed formation rule, a shall-operator operates on only one verbal noun. Thus the conjunctive statement16

Tom will shortly cross the road and Dick will shortly sit down

gives rise to

Shall [its being the case that Tom will shortly cross the road and Dick will shortly sit down].

This means that for simplicity of representation we can leave the job of turning indicative statements into verbal nouns to the brackets and write the shall-statements corresponding to a given indicative as

Shall [(indicative)]

Since we are representing shall-statements as a matter of an operator operating on a gerund, it might seem appropriate to follow Hare by interpreting indicative statements as a matter of an operator operating on a gerund also. This, however, is not only unnecessary, but seriously mistaken. Let us consider the indicative counterpart of the practical reasoning about boxes dissected above.

M'(1) This is one of the boxes

So, (Jones will shortly take all the boxes to the station) dependently implies (Jones will shortly take this box to the station)

M'(2) Jones will shortly take all the boxes to the station

So, Jones will shortly take this box to the station.

The important thing to note is that the "implies" of M'(1) is a relation word which takes singular terms for its arguments. In the present context we are using it to take verbal nouns for its arguments. In other, but related, uses "implies" takes that-clauses or quoted expressions for its arguments. These singular terms are derivative from the corresponding indicative statements. One operates on an indicative statement to get the corresponding singular term. Thus in

(S is P) implies (S is Q)

the indicative statements "S is P" and "S is Q" have been turned into the singular terms "(S is P)" and "(S is Q)." Thus an inference in accordance with this license is to be understood not as

S being P, yes
So, S being Q, yes

which adds a yes-operator to a "phrastic," but rather as simply

S is P
So, S is Q

which subtracts the singular term operator which "suspends" "S is P" and "S is Q" in the inference ticket. Correspondingly, M'(2) is not to be "reconstructed" as

(Jones's being about to take all the boxes to the station) yes
So, (Jones's being about to take this box to the station) yes

for it is "perspicuous" as it stands.

I have chosen my examples and oriented my discussion so as to imply rhat "shall" inference tickets are always of the form

(Shall [--]) implies (shall [. . .])

and rest on a corresponding indicative inference of the form

(---) implies (. . .)

Whether or not this is always true will be discussed at a later stage in the argument.

5

To sum up the argument of the past few sections:
(1) There is no such thing as imperative inference, i.e., inference involving tellings to as tellings to. There is, however, practical reasoning, and there is argument involving tellings of intentions.
(2) We have emphasized that the only sense in which there is a special logic of imperatives is that exhibited by the reasoning

X tells Y (to do A if p)
p
So, X tells Y (to do A)

which is the counterpart of

X tells Y (that q if p)
p
So, X tells Y that q.

That these arguments are valid (with some requirement as to the ascertainability of p) is analytic of the concept of telling as a performance. Note that while

X tells Y to do A
Doing A entails doing B
So, X tells Y to do B

and

X tells Y that p
That p entails that q
So, X tells Y that q

are valid arguments, the same is not true if a less restricted use of "implies" is substituted for "entails." Indeed, it can be argued that this is the locus of the difference between what statements entail and what they imply without entailing. A related point concerns the validity of the arguments

X intends Y to do A *
Doing A entails doing B
So, X intends Y to do B

and

X believes that p
That p entails that q
So, X believes that q

But to follow up the point would take us too far afield.
(3) We have suggested that "shall" inference tickets have the form

(Shall [---]) implies (shall [. . .])

where the shall-operator appears in both antecedent and consequent, and that they are parasitical upon the indicative inference tickets of the form

(---) implies (. . .)

where the square brackets of the "shall-" tickets are understood to make participial expressions of the statements mentioned by the indicative inference ticket. It is important to note, though the full significance of the fact lies outside the scope of this essay, that although in the context "(---) implies (. . .)" gerund expressions are to be construed as singular terms, in the simple context "shall [---]" they are not. In other words, shall-statements, unlike implication statements, are in the object-language. In this respect "shall" resembles truth-functional connectives.

Universality and the Logic of Intentions
6

It will be useful to conclude these animadversions on the logic of imperatives and resolutives with some remarks on the hypothetical imperative.17 My main point is that to say "If you want A, do B!" is not a special case of telling someone to do something. Here, again, we find the imperative mood serving as the vehicle of a performance, but the performance is that of giving advice, rather than telling someone to do something.18 The "want" of "If you want A, do B!" like the "want" of "What does he want me to do?" is roughly equivalent to "intend."19

It might seem plausible to interpret the hypothetical imperative

If you want the red one, take that!

uttered by Jones to Smith, as having the force of

(Since that is the red one and you (Smith) want the red one).
''You (Smith) take the one you want!" implies "You (Smith) take that!"

and hence as the principle of the argument

You (Smith) take the one you want!
The one you want is the red one
That one is the red one
So, you (Smith) take it.

But, as we have already pointed out, the hypothetical imperative belongs to the category of advice, and in advice the imperative mood is used in answer to the question,

What shall I (Smith) do?

It is, therefore, concerned with the questioner's intentions, whereas the telling to use of imperatives answers the question,

What do you want me to do?

and expresses the intention of the person who uses the imperative. When Jones advises Smith by saying "Do A!" in answer to the question "What shall I do?" Smith accepts this advice by forming the intention expressed by "I shall do A." It is as though Smith handed to Jones the incomplete sentence,

I (Smith) shall do . . .

and Jones had added "A" to complete it.

In the case of the hypothetical imperative, "If you want the red one, take that!" Jones is telling Smith that that is the red one. He is, however, giving this information in a way which reflects the deliberative role it would play in Smith's thinking if, indeed, he does intend to take the red one. This role can be represented by

(1)

That is the red one
So (I take the red one) implies (I take that)
So, (Shall [I take the red one]) implies (Shall [I take that])

(2)

Shall [I take the red one]
So, shall [I take _that_]

Thus, to advise Smith with the hypothetical imperative in question is to answer, by anticipation, the question

I (Smith) shall take the red one, so which shall I take?

The answer, however, though verbally similar to the telling to "You (Smith) take that!" conveys the information that that is the red one for use as a premise in Smith's reasoning as outlined above.

7

Let us now look at the implications of the above considerations for the analysis of ought-statements. At the end of Section 4 we were preparing to examine the idea that ought-statements are truly universal imperatives; that they have a force which it takes what is, from the standpoint of actual usage, the absurdity "Everybody do A in C throughout all past and future time!" to represent. We are now in a position to understand this absurdity and to see that it just won't do to suppose that by a simple extension (for analytic purposes) of the ordinary imperative mood, we can represent ought-statements as truly universal imperatives. Reflection on the fact that imperative utterances are practical performances to which appeal can be made in justifying or excusing our conduct, makes it clear that they presuppose publication to those who are to do what the imperatives tell them to do. It is no accidental feature of imperatives that one can only tell people to do things in the future. And it is no accidental feature of imperatives that they are formulated by the use of tensed verbs in which the tense has its full and ordinary force. There can be no counterpart here of the "tenseless" is which philosophers use to formulate truly universal matters of fact. Only God at the time of creating Adam could have sensibly used the imperative "Everybody at all times and places do A in C!" for only then was all relevant time all future time.

But if we cannot tell past people to do A in C, we can wish that they had done so. If moral principles cannot be interpreted as truly universal imperatives, may they not be the expression of truly universal wishes? "Everybody ought to do A in C," then, would have the force of "Would that everybody had done A in C, and did A in C from now on" -- a wish which we may abbreviate to

Would that everybody universally did A in C!

This suggestion has the advantage (over the imperativist account) that

Would that everybody universally did A in C!
X is (was, will be) in C
So, would that X did (had done) A!

has the feel of deliberative reasoning. And though there are fairly obvious reasons to mistrust this suggestion, let us refuse, for the moment, to entertain them, for it will be useful to let it grow before it is modified or abandoned.

Let us suppose, then, as a working hypothesis, that

Everybody ought to do A in C
X doing B in these circumstances is X doing A in C
So, X ought to do B

has the force of

Would that everybody universally did A in C!
X doing B in these circumstances is X doing A in C
So, would that X did B!

and let us suppose, again as a working hypothesis, that

Would that I did B

has the force of

I shall do B

and

Would that X did B!

the force of

X shall do B

where the shall-sentences are used in accordance with the stipulations of Section 3. Now let us add the logical point, which derives from the analysis of what it is to have an intention, that, other things being equal, a person who candidly says

I shall do B

and is not a victim of self-deception, and is not mistaken about the circumstances, and does not change his mind, will do B; and that, with similar qualification, one who candidly says

Tom shall do B

will do that which he believes would bring about Tom's doing of B. If we add this logical point, can we not claim that the above is at least a first approximation to an analysis of what it is to have a moral principle and act on it?

8

But if this account has the merit of freeing ought-statements from "telling to," while preserving the connection with "intending" which was the sound core of the imperativist analysis, it won't do as it stands. Much more remains to be said about the universality represented, in this account, by "Would that everybody universally did A in C!" Then, when and if this hurdle has been crossed, the account must be freed from its stress on sentences having the special force of the form "Would that X did A!" Of these tasks, the former will prove decisive.

Consider the case of someone who has a universal wish in a more ordinary and restricted sense of "universal." Suppose that Jones wishes that "everybody" do a certain kind of action, where the scope of "everybody" is, say, the people who are with him on a particular occasion. We who reflect on Jones's intention may ask ourselves

Why (for what reason) does Jones wish that everybody do A? and we may arrive at an answer of one of the following types:

Because if everybody does A, this would have as its joint result the state of affairs S (which he wishes to exist for its own sake).

Because each doing of A has a consequence of a certain kind K (which he wishes to exist for its own sake), and if everybody does A, this would bring about the maximum number of K states of affairs.

Because if everybody does A, everybody does A (i.e., everybody doing A is a state of affairs which he wishes to exist for its own sake; he is interested in single doings of A only as logically necessary conditions of everybody doing A.

Because if everybody does A, there are that many individual doings of A. (He would wish any A-doing to exist for its own sake). I do not wish to imply that this list exhausts the possible types of answer to the above question. There is also the answer

Because they ought (as he thinks) to do A. This answer, however, only poses all over again the puzzles we are seeking to solve.

Answers of type (a) and (b) would seem to be clearly irrelevant to the analysis of action on principle, and I shall just assume, for the time being, that there is no sense in thinking that having the principle Everybody plight to do A in C is a matter of wishing that everybody did A in C for the sake of the consequence of such action. Answers (c) and (d) are useful points of departure for the solution of our first problem, that concerning the universality of ought-statements. Let us consider them in reverse order.

The first thing to note is that while (d) grants that Jones wishes that everybody did A, it emphasizes that it is individual cases of A-doing in which he is interested; his wish that a given case of A-doing exist is not a wish that it exist in order that a set of A-doings exist. Rather, his wish that everybody do A sums up, so to speak, his wishes that this, that, and the other person do A. He "just likes" individual cases of A-domg to exist, the more the merrier. He does not deliberate

Would that "everybody" did A!
So, would that Tom did A!

If anything, he deliberates

Would that Tom, Dick, Harry, indeed that "everybody" did A!

Case (c), on the other hand, provides us with an example of

Would that everybody did A!
So, would that Tom did A!

Yet it clearly does not give us what we are looking for, as it specifies that Jones is interested in Tom's A-doing only as a necessary condition of the state of affairs everybody doing A, that is to say, as one component in the state of affairs

Tom doing A and Dick doing A and. . . .

But action on principle is not silly, as it would be if it were a matter of wishing that I keep this promise as a logically necessary condition of a world-long and world-wide keeping of promises.

If only we could combine that feature of (c) which is expressed by

Would that "everybody" did A!
So, would that Tom did A!

with that feature of (d) which is expressed by

Whether or not the others do A, would that Tom did A!

Now the moral of the above example is that to understand action on principle, it is not enough to invoke separately the ideas of universality and not as a means. The universality of the intention of action on principle is the manner in which the action is not intended as a means. To understand the universality of the intention in action on principle is to understand the sense in which a particular action, done on principle, is done for the sake of the action itself.

The point at which I am driving is, perhaps, best brought out by making a somewhat parallel point in another context, though to make it in such brief compass I must skate hurriedly over thin ice. To acknowledge that

All M's are universally N's

is to be prepared candidly to say such things as

This is an M, therefore it is an N
That is an N because it is an M
If this were an M, it would be an N
If this had been an M, it would have been an N
If anything were an M, it would be an N.

The point of the parallel is that

Would that everybody universally did A in C!

is the expression of commitment to a principle, only if it expresses a readiness to such reasonings as

Would that Tom did A, for he is a person in C
I am a person in C, so would that I did A!
Since Dick was a person in C, would that he had done A!

Once again, let us permit the suggestion to grow before submitting it to a closer scrutiny. If it is sound, then to represent the deliberation which culminates in "I ought to do this" by the form

Would that everybody universally did A in C!
I am in C
So, would that I did A!

would be to make the mistake of treating an expression of the principle in accordance with which one reasons about particular cases as though it were the major premise of such reasonings. The corresponding mistake in the case of theoretical (as contrasted with practical reasoning) is represented by the form

All M is necessarily N
This is M
So, it is N

We shall come back to this point in a moment. In the meantime we can see why, if the suggestion is sound, approving on principle of a doing of A in C is not the same thing at all as favoring it simply because one recognizes it to be a doing of A in C. To say of Jones that he favors an item simply because he thinks of it as being of a certain description is to say that it is the fact that Jones thinks of the item as being of that sort which explains why he favors it. We can represent this by the reasoning

Jones thinks that x is f
So, Jones approves of x

and the inference ticket which authorizes this reasoning as

Whenever Jones thinks that an item is f, he approves of it.

But the fact that we reason about Jones's approval of x along these lines must not be confused with the idea that Jones's approval of x is a reasoned approval. That is, we must avoid the assumption that because we can correctly argue as above, Jones must have reasoned

x is f
So, would that x were defended, etc.

It is simply not true that if Jones's thought that x is f is the explanation of the approval, then his approval must be a _reasoned_approval. It may be, or it may not. And if it is a reasoned approval, then the explanation of his approval of x is, strictly speaking, not

Jones approves of x because he thinks that it is f

but rather,

Jones approves of x because he thinks that x is f and accepts the practical principle "If anything is f, would that it be defended, etc."

To represent Jones as thinking

Would that x, which is f, were defended, etc.

is not the same thing at all as representing him as reasoning

x is f
So, would that x were defended, etc.

even though we add to the former the information that whenever Jones thinks of an item as f, he thinks (ceteris paribus)

Would that it were defended, etc.

As a parallel it may be pointed out that (as Prichard has emphasized)

Jones thought that it would thunder because he thought that lightning had just occurred is authorized by "Whenever Jones thinks that lightning has occurred, he thinks that it will thunder," and can be a mere matter of the "association of ideas," whereas

Jones thought that it would thunder because lightning had just occurred (Jones thought "It will thunder because lightning has just occurred") cannot. A psychological commentary on Jones's thinking in the latter case would have to mention Jones's acceptance of the principle "One may infer the occurrence of thunder from that of lightning" (Jones thinks "Whenever lightning occurs it will thunder"). Notice that (2) endorses the idea that lightning has just occurred. If the speaker does not wish to endorse Jones's thought that lightning has just occurred, he would say

Jones thought that it would thunder because (he thought) lightning had just occurred. This parenthetical use of "he thought" must not be confused with the "he thought" of "because he thought that lightning had just occurred."

Thus, we see, it is essential not to confuse

Smith did B because he thought his doing B would be doing A in C

with

Smith did B because, he thought, his doing B would be doing A in C.

The difference in punctuation highlights the difference between asserting that the occurrence of thoughts of the form "my doing B would be a case of doing A in C" constitutes a (partial) explanation of the occurrence which was Smith's doing of B, and asserting that it was practical thinking of the form "I shall do B because my doing B would be a case of doing A in C" which accounts for Smith's doing of B. In the second case, the "he thought" is a parenthetical comment by the speaker on Smith's reason for doing as he did. It must, of course, be granted that reasons are causes, i.e., that in general we can move from statements of the second form to statements of the first, and vice versa. But these moves are not without their dangers. For, as is well known, in causal explanations we are content to single out one aspect of the total relevant situation within which the explanandum occurs, dub it the cause, and relegate the presupposed remainder to the category of condition. On the other hand, we have not given a person's reason for acting as he did, unless we have given the whole reason in its proper form. Thus, in attempting to indicate that Smith has acted on principle, we may begin by saying, in the order of causes

Smith did B because he thought that to do B in his circumstances would be to do (an action of kind) A in (circumstances of kind) C.

This, however, leaves it open whether, in our statement of Smith's reason for acting as he did, we should say

Smith did B because, he thought, doing B in his circumstances would be doing A in C or

Smith did B because, he thought, doing B in his circumstances would be a case of anybody doing A when in C. It is only (B) which represents Smith as reasoning (roughly),

(B') Would that I did A, if I am in C
(In the case of anybody, would that he did A when in C)
To do B in these circumstances is to do A in C
So, would that I did A.

On the other hand, (A) is satisfied if Smith's decision is of the form,

(A') Would that I did A, if I am in C
To do B in these circumstances is to do A in C
So, would that I did B

provided only that if Smith were to think of any action of any person as a doing of A in C, he would wish it done for its own sake, and hence reason in each case

(A'') Would that Y did A, if Y is in C
To do B in Y's circumstances is to do A in C
So, would that Y did A

Even with this proviso, however, (A) does not acquire the force of (B). To reason in each case as in (A'') is not the same thing as to reason in each case in terms of a principle applicable to each case. Yet we will not have clarified the difference, in the case of practical reasoning, until we understand the status of the bracketed move in reasoning (B) above.

Notice, also, that if this proviso is not added, and if, for example, the doing of A in C by other people would not be approved by Smith, then we would have to revise the statement of Smith's reason for doing what he did to read, correspondingly,

(A''') Smith did B because, he thought, doing B in his circumstances would be a doing of A in C by Smith

and the original causal statement to read

Smith did B because he thought doing B in his circumstances would be a case of Smith doing A in C.

Approving on principle is not the same as being disposed to "just like" each item which one comes to think of as being of a certain kind.20 This is not, for the moment at least, to be construed as a denial that people might arrive at the espousal of a principle of doing A when in C by a process which began with "just liking" (however this might come about) individual cases of people doing A in C; nor that, regardless of how we come to espouse it, the principle, when challenged, might be "justified" by the fact that when all the chips are down and all the information in, we find that we "just like" any case of doing A in C, in a sense of "just like" which is not the espousal of the principle all over again.

9

Let us now return to the topic of universality. We had arrived at the point of suggesting that "Would that everybody universally did A in C!" expresses the espousal of a principle of conduct, as expressing a readiness to reason in accordance with the form

X is (was, will be) a person in C
So, would that X did (had done) A

where X is a variable which ranges over persons, that is to say, us.21 But before we proceed, we must, as already noted, free our account from its overly close connection with wishing, which stems, it will be remembered, from its genesis in a critique of the imperativist analysis. But instead of examining the logic of "to wish" and showing how "would that . . ." is related to other forms of practical discourse in order to show how wishes can embody the same principles as other practical "attitudes," I shall limit myself to pointing out that while neither "X should have done A" nor "It should have been the case that p" can be said to express intentions, any more than "X did A" or "It was the case that p" can be said to express expectations; nevertheless, intending may have its past tense (and subjunctive) counterparts, as expecting has its counterpart in historical thinking. A person who thinks at 10:00 A.M.:

Shall [X doing A ten minutes from now]

can be said to intend that X do A in ten minutes. Is there no practical thought that he can think about X doing A at 10:20, assuming that he has not changed his mind? It is surely plausible to interpret

Should [X doing A ten minutes ago]

as a differently tensed counterpart of the former. But what role might such thoughts play? It is worth noting that "X should have done A" is much closer to "X ought to have done A" than are simple "shalls" to "oughts." If we throw more light on the relation of "ought" to "shall," we may then be in a position to appreciate practical discourse in the historical mode. But first let us continue with our attempt to build "shall" into "ought" without letting that problem distract us.

At the end of Section 4 we argued that reasoned decisions involving "shall" were ultimately of the form

Shall [---]
So, shall [. . .]

where the implication authorizing this inference, namely,

(Shall [---]) implies (shall [. . .])

rests on an implication between the indicative statements "---" and ". . ." which, by being bracketed, i.e., turned into participial expressions, serve as operands which the shall-operator turns into shall-statements. A not too complex example of practical reasoning of this kind would be

I shall do A, if I have the money
I have the money
So, I shall do A

which, on the above analysis, breaks down as follows:

Shall [my doing A, if I have the money]

(I have the money) implies ([I will do A, if I have the money] implies [I will do A])

I have the money

So, (my doing A, if I have the money) implies (my doing A)

(Shall [my doing A, if I have the money]) implies (Shall [my doing A])

So, Shall [my doing A] In this "breakdown" of the original inference, the antecedents and consequents of implication statements (2), (4), and (5) are either both indicatives or both resolutives. In other words, whereas the original formulation suggests that the reasoning proceeds in accordance with the principle

(I have the money) implies (Shall [my doing A])

or, perhaps,

"(If I have the money, then shall [my doing A]) and I have the money)" implies (Shall [my doing A])

in neither of which is the antecedent a shall-statement, the proposed breakdown separates the reasoning into separate stages which are either pervasively indicative or pervasively resolutive.

Against this background what are we to make of the reasoning, characteristic of action on principle,

Jones is a person in C
So, Jones shall do A

Our first attempt may well be to construe this reasoning as authorized bv the general inference ticket

(X is a person in C) implies (X shall do A)

But as this ticket has a shall in its consequent, but no shall in its antecedent, we may be tempted to think of it as a derivative inference ticket, one which must be explicated in terms of an inference ticket moving from shall to shall, and which, in its turn, rests on an inference ticket which moves from indicative to indicative. Thus, we might be tempted to reconstruct the argument as follows:

Shall [(X) (X doing A, if X is in C)]

"(X) (X does A, if X is in C)" implies (Jones does A, if Jones is in C)

"Shall [(X) (X doing A, if X is in C)]" implies (Shall [Jones doing A, if Jones is in C])

Shall [Jones doing A, if Jones is in C]

(Jones is in C) implies ([Jones doing A, if Jones is in C] implies [Jones doing A])

Jones is in C

(Jones doing A, if Jones is in C) implies (Jones doing A)

(Shall [Jones doing A, if Jones is in C]) implies (Shall [Jones doing A])

Shall [Jones doing A] But this suggestion, according to which moral principles are universal resolutives of the form

Shall [(X) X doing A, if X is a person in C]

runs up against the objection that it would be silly to espouse such resolutives knowing full well that in a great many cases people who are in C simply have not done, nor will they do, A. A more plausible approach, which has the advantage of interpreting the universality of moral principles in terms of each case rather than the totality of cases, draws on the concept of an axiom schema. According to it, instead of saying that

Shall [(X) (X doing A, if X is a person in C)]

is an axiom, we should rather say that every statement derived from the schema

Shall [X doing A, if X is a person in C]

by replacing X with the name of a person is an axiom or first principle of moral reasoning. According to this approach, the first three steps of the above breakdown are to be replaced by the characterization of (4) as an axiom conforming to the above axiom schema.

But what is the difference between the following two conceptions of espousing a moral principal pertaining to people doing A in C?

to espouse such a moral principle is to accept as a first principle of practical reasoning any resolutive conforming to the schema

Shall [X doing A, if X is in C]

to espouse such a moral principle is to accept the general inference ticket,

(X is a person in C) implies (Shall [X doing A]) To choose between these interpretations requires a closer look at conditional intentions, i.e., at the logical form of "X shall do A, if p" and its relation to the inference

X shall do A, if p
p
So, X shall do A.

It has often been argued that the closest that ordinary language comes to exhibiting the truth-functional connective of material implication in its pure form is in the expression of conditional intentions. Or, to put the point the other way around, that whereas most ordinary uses of "if . . . , then --- " give expression to basic or derivative relations of inferability, the "if . . . , then ---" of conditional intention has nothing to do with inferability, but rather is material implication pure and simple. But this way of putting the point is too strong; for if a person says

I shall do A, if p

and we subsequently find that in spite of becoming convinced that p, he does not do A, we shall infer either that he has changed his mind, or that he was mistaken about his frame of mind, or that he was deceiving us. Thus there is a logical connection between "I shall do A" and "p" stronger than material implication, and it would be incorrect to say that the force of "I shall do A, if p'' has nothing to do with inferability. On the other hand, the inferability seems clearly to relate to the force of "shall" as governing the entire conditional, and not only to be compatible with the absence of an independent intentional connection between p and the doing of A but also to require that there be none. The inferability is exhibited by the following schema, in which (and hereafter in this essay) "if .... then --- " stands for material implication.

X candidly says (thinks) "Shall [if p, then my doing A]"
X candidly says (thinks) "p"
So, X will do A

If so, then it is a mistake to interpret the "if . . . , then --- " of conditional intention as anything more than material implication, even though one must grant that a relation stronger than material implication obtains between the corresponding biographical statements concerning intentions, beliefs, and actions.

These considerations throw light on the difference between interpretations (A) and (B) above of what it is to espouse a moral principle. For it enables us to dispel the feeling that these proposals are equivalent. This feeling rested on the rough equivalence between accepting the inference ticket

(x is f) implies (x is g)

and accepting the schema

If x is f, then x is g

as defining a class of axioms (i.e., unconditionally assertable statements). Thus, if to espouse a moral principle were to accept the schema

If X is in C, then X shall do A

as defining a set of axioms, one might well conclude that to speak instead of accepting the inference ticket

(X is in C) implies (X shall do A)

would be to say essentially the same thing in other words. But if, as we have argued, it is a mistake to interpret expressions of conditional intention as having the form

If p, then X shall do A

and if they must, instead, be attributed the form

(Shall [if p, then X doing A]), i.e., (Shall [if p, then X doing A])

as must be done to bring the condition within the intention, the parallel disappears. Thus, (B) must be rejected, and (A) so interpreted as not to confuse

X accepts as a first principle of practical reasoning any resolutive conforming to the schema: (Shall [if X is in C, then X doing A])

which is (in first approximation) the correct account, with

X accepts as a first principle of practical reasoning any resolutive conforming to the schema (if X is in C, then shall [X doing A])

which is not.

This account of espousing a principle has the additional merit of making clear why it won't do to interpret such espousal as a matter of having a general intention, thus:

X accepts as a first principle of practical reasoning the universal resolutive, "Shall [(X) (if X is in C, then X doing A)]."

For to intend in this sense that everybody do A in C, knowing, as one does, that "(X) (if X is in C, then X does A)" is false, would be, if not (as it seems to be) logically impossible, at least silly.

Let me now suggest as the next step in the analysis of ought-statements that

X (as being a person in C) ought to do A

has the force of

Shall [X doing A] because X is in C and shall* [if X is in C,then X doing A]

where the asterisk attached to the second "shall" is a signal that the speaker recognizes resolutives of the form

Shall [if X is in C, then X doing A]

as first principles of practical reasoning. Needless to say. this answer is oversimplified in a number of respects, not the least of which is its presupposition that the A and C of the ought-statement are the action and circumstance categories of a principle at hand. For the relation between ought-statements and principles is more flexible than indicated above. To make an ought-statement one need not, as we saw in Section 4, have a principle up one's sleeve; rather, one commits oneself to the idea that "there is" an explanation having the form of the above because-statement, X's circumstances being of kind C and the action in question of kind A.

A more radical defect in the above "analytical model," as we shall see. is its neglect of that dimension of moral principles which is the fact that they are our principles (not merely my principles) though it does justice to the fact that they are principles about us.

10

There is a consideration pertaining to intentions and their expression which, though not strictly a part of the argument of this paper, indicates how it might fit into the broader framework, of an empiricist philosophy of mind.22 It Is that to intend that person P do A is to think "P shall do A." i.e., to be disposed to have thoughts of the kind "P shall do A" where these thoughts are inner episodes construed on the model of the overt utterances which, in candid discourse, are initiated by these inner episodes, and in this sense, express them. (To believe that x is f is, correspondingly, to think "x is f," i.e., to be disposed to have thoughts of the kind "x is f," where these thoughts are inner episodes construed on the model of the overt utterances which, in candid discourse, are initiated by these inner episodes, and would be said to express this belief.) And if we must add that in a certain sense one can think "P shall do A" without really intending that P do A, this is not because intending that P do A is thinking "P shall do A'' plus something else, but because that which appears to introspection as the thought "P shall do A" is only presumptively this thought. Thus, if someone candidly assures us "I shall do A," and it turns out at the appropriate time that, although he assures us that he has not changed his mind, the thought of doing A has no power to move him to act, then it would be proper to deny not only that he had really intended to do A, but also that the presumptive thought "I shall do A." which he had introspected, really was such.

Notice that I am not saying that intending, for example, to do B in order to bring about A, is identical with the power of the thought "My doing B would bring about A" to move one to act. I am, however, insisting that the power, in the causal order, of thoughts of the form "My doing B would bring about A" to move me to act is a presupposition of the language of "shall" and the order of reasons -- indeed, that "I shall do B in order to bring about A" embodies this presupposition.23

It follows from the above conception of the status of intentions and beliefs, together with the interpretation of ought defended in this essay, that to believe "P ought to do A" is, in part, to intend "P shall do A."

11

We must now face up to the fact that "shall" and its kindred express the intentions of the speaker. Thus, and the point is an obvious one,

Jones shall do A

differs from

Jones intends to do A

in that whereas the speaker uses the latter to ascribe to Jones the intention to do A, the former expresses the speaker's intention that Jones do A. Thus, if I say to Jones

You shall do A

I have no logical right to expect him to concur by saying

Yes, I shall do A.

And if he should reply

No, I shall not do A

the logic of "shall" makes no demand that one or other of us change our mind by abandoning our intention -- though logic does assure us that these intentions are incompatible in that they cannot both be realized. Consequently, if my statement

You (Jones) ought to do A

were simply

You (Jones) shall* do A

where the asterisk signals the presupposition of practical reasoning of the form

Shall [if Jones is in C, then Jones doing A]
So, Jones being in C, shall [Jones doing A]

in which the major premise is an axiom of the type discussed in Section 9, there would be no logical reason to expect Jones, given that he agrees that he is indeed a person in C, to meet my

You (Jones) ought to do A

with

Yes, I (Jones) ought to do A.

My point is that the language of morals differs from the language which simply expresses the speaker's intentions, not only by expressing or implying reasoning of the kind we have been exploring, but also by presupposing (in one sense of this much abused expression) that all disagreements of the form

A: Jones ought to do one thing
B: No! He ought to do another

are "in principle" reducible to disagreement about matters of fact and not to disagreement in intention.

How is this presupposition to be understood? The explanation centers around a simple fact about the grammar of the word "ought": that ought, unlike shall, has a proper negative. Whereas shall is characterized by the two forms

Shall [X doing A]
Shall [X not doing A]

ought enjoys the full complement

Ought [X doing A]
Ought [X not doing A]
Not-ought [X doing A]
Not-ought [X not doing A]

In short, one person can contradict another person's ought, whereas shalls conflict but do not contradict.

Let us introduce the concept of the biographical counterpart of a shall-statement. Thus, corresponding to Tom's shall-statement

Shall [X doing A]

there is the biographical counterpart

Tom intends that X do A

which, given that Tom speaks candidly and without self-deception, will be true. Using this concept we can pair off the following shall-statements and biographical counterparts:

(1a) Tom intends that X do A
(1b) Tom: ShaU [X doing A]
(2a) Tom intends that X not do A
(2b) Tom: Shall [X not doing A]
(3a) Tom: intends that X do A or not do A as X pleases
(3b) Tom: Shall [X doing A, if X so wishes and X not doing A, if X so wishes]
(4a) Tom has formed no intention with respect to X doing A
(4b) Tom: Shall [X doing A]?24

It is clear that these relationships do not require the form

Not-shall [X doing A]

because the one case which might seem to call for it, namely (3a), simply gives us another example of a shall-statement. But what of biographical statements of the form "Tom does not intend that X do A"? Do we not have

(5a) Tom does not intend that X do A
(5b) Tom: Not-shall [X doing A]

and is not (5b) equivalent to

(5c) Tom: May [X not doing A]

The answer is that (5a) is compatible with (4a), whereas (5c) clearly expresses the outcome of a deliberation, and is, indeed, the expression of an intention. In fact, it would not be implausible to regard (5c) as a variant form of (3b).

Against the equation of (5c) with (3b), however, there are important considerations pertaining to the role of "may" in telling to discourse. It is, I believe, illuminating to regard the fundamental role of "may" as that of withdrawing a telline to. We have already called attention to the difference between expressing the intention that X do A, and telling X to do A, a difference which is represented in our symbolism by the difference between "Shall [X doing A]" and "Shall [X doing A]!" If, now we pair off tellings to with their biographical counterparts, we have, for example,

(6a) Tom tells X to do A
(6b) Tom: Shall [X doing A]!

Here the fact emerges that if Tom has told X to do A, he may subsequently (subject, of course, to certain conventions) withdraw this performance by saying "May [X not doing A]!" This direct withdrawal is not to be confused with the "implicit" withdrawal which is performed by "contramanding,"25 i.e., presenting X with a new telling to, one which conflicts with the old, as,

Tom: Shall [X doing A]!
Tom (subsequently): Shall [X doing B]!

where doing B entails not doing A; or, in a weaker form,

Tom: Shall [X doing A]!
Tom (subsequently): Shall [X not doing A, if p]!

Withdrawing a telling to is the practical counterpart of withdrawing a telling that. Therefore, it is important to distinguish between withdrawing a telling that, and contradicting a telling that. The difference is illustrated by the dialog,

Tom: S is P
Tom (subsequently): S is not P
Dick (chiming in): S is not P

Tom has both contradicted and by implication withdrawn his earlier telling that. Dick obviously cannot withdraw Tom's telling that; but he can, and does, contradict it. Notice that Tom might have withdrawn his earlier telling that by saying "S may not be P after all." This use of "may" is not to be confused with notions pertaining to possibility or probability in those uses w'hich do have proper contradictories.

The fallacy I am attempting to expose is exhibited by the following sequence of statements:

Countermanding is the counterpart of contradicting

To contradict is to assert the negative of what was asserted

"May [X not doing A]!" countermands "Shall [X doing A]!"

"May [X not doing A]!" = "Not-shall [X doing A]!" The fallacy lies in the third step. "May [X not doing A]!" countermands "Shall [X doing A]!" in a sense which is the counterpart of contradiction, only if it is interpreted, roughly, as saying "Shall [X doing A, if X so wishes and X not doing A, if X so wishes]!"; in which case it is not a "not-shall" statement (whatever that might be) but simply another shall-statement which implicitly withdraws the original telling to. If, on the other hand, "May [X not doing A]" is taken to play the direct withdrawal role, then it is not the counterpart of a contradiction, and the argument also breaks down.

Universality and Intention: A Second Mode
12

I pointed out above that whereas one person can contradict another person's ought, shalls conflict but do not contradict. But to make it intelligible how ought can "at bottom" be a shall. and yet have this radical inrersubjectivity, we must bite a bit deeper. Consider the following exchange:

D(1)

Tom: S is P

Dick: S is not P

Dick has contradicted Tom; and supposing candor on all sides, we can correlate with this dialog the following biographical counterparts:

BD(1)

Tom thinks that S is P

Dick thinks that S is not P

Consider, next, the dialog,

D(2)

Tom: I (Tom) shall do A

Dick: You (Tom) shall not do A

This time the biographical counterparts are

BD(2)

Tom intends that Tom do A

Dick intends that Tom not do A.

If we reflect on these two pairs, we notice a point of resemblance and a point of difference. The point of resemblance is that in each of the dialogs what each speaker says expresses his frame of mind -- in the one case the thought that S is (or is not) P, in the other the intention that Tom do (or not do) A. The difference consists in the already noted fact that whereas in the first dialog Dick has contradicted Tom, in the second he has simply expressed an intention which is not co-realizable with Tom's intention.

Consider, next, the two dialogs,

D(3)

Tom: I (Tom) ought to do A

Dick: You (Tom) ought not to do A

D(4)

Tom: I (Tom) ought to do A

Dick: It is not the case that you (Tom) ought to do A.

Our analysis suggests the following biographical counterparts for the first of these dialogs,

BD(3)

Tom intends that Tom do A because he has such and such an "axiomatic" intention and such and such factual beliefs

Dick intends that Tom not do A because he (Dick) has such and such an "axiomatic" intention (about Tom) and such and such factual beliefs.

But it leaves us puzzled as to just what to offer as the counterpart of D(4). For if our aim is to reconcile the idea that in D(4) Dick is contradicting Tom with the idea that the ought is a special case of shall, we can scarcely be satisfied wirh

BD(4)

Tom intends that Tom do A

It is not the case that Dick intends that Tom do A.

For whereas in the biographical counterpart of D(1) the contradictory statements "S is P" and "S is not P" reappear in the guise of that-clauses; in BD(4) the supposedly contradictory statements have simply disappeared. On the other hand, if we simply represent the biographical counterpart of D(4) as

BD(4')

Tom thinks that Tom ought to do A

Dick thinks that not-(Tom ought to do A)

we say what is true, but lose contact with the "analysis" of ought in terms of shall.

I suggest that (to put it in a radically oversimplified manner) the autobiographical counterpart of an ought-statement is not simply

I intend that X do A because I have such and such an "axiomatic" intention with respect to X, and such and such factual beliefs

but rather

We intend that X do A. . . .

Thus, corresponding to the statements,

D(5)

Tom: I (Tom) ought to do A as being a person in C

Dick: You (Tom) ought to do A as being a person in C

Harry: He (Tom) ought to do A as being a person in C

we have as counterparts not

BD(5)

Tom "axiomatically intends" that Tom do A as being a person in C

Dick "axiomatically intends" that Tom do A as being a person in C

Harry "axiomatically intends" that Tom do A as being a person in C

but, rather, the autobiographical counterparts,

ABD(5)

Tom: We "axiomatically intend" that Tom do A . . .

Dick: We "axiomatically intend" that Tom do A . . .

Harry: We "axiomatically intend" that Tom do A. . . .

And if, with this in mind, we form the autobiographical counterpart of D(4), we have something like,

ABD(4)

Tom: We "axiomatically intend' that Tom do A ...

Dick: We do not "axiomatically intend" that Tom do A. ...

What is the force of such phrases as "we intend . . ." and "we are committed to the intention . . ."? Here we touch on the "institutional" aspect of morality. For "We intend . . ." is clearly not the logical sum, so to speak of "Tom intends . . . ," "Dick intends . . . ," "Harry intends . . . ," etc. Nor does

Tom (who is one of us) does not intend that X do A

contradict

We intend that X do A

any more than

Tom (who is one of us) doesn't mind women smoking

contradicts

We disapprove of women smoking.

Nor (I need scarcely add) does the fact that this is so involve the existence of a "group mind," capable of having beliefs and intentions, in a sense incompatible with empiricist principles. Empiricism has properly stressed the logical dependence of concepts pertaining to the beliefs and attitudes of groups on the corresponding concepts pertaining to individuals. This dependence, however, as the above examples make clear, involves a certain flexibility. Nevertheless, the fewer the people in the group who believe that p or intend that X do A, the less defensible becomes the statement that the group believes that p or intends that X do A. These are familiar considerations. I wish to emphasize that when the concept of a group is "internalized" as the concept of us, it becomes a form of consciousness and, in particular, a form of intending.

We saw above that "We intend that X do A" does not entail "Tom (who is one of us) intends that X do A." On the other hand, it is clear that a person who shares none of the intentions of the group could scarcely be said to be one of us. There is a particularly close logical connection between "We intend . . ." and "I intend. . . ." This does not mean that it can never be proper to say "We intend . . . , though I, for myself, do not." On the contrary, this makes perfectly good sense. Yet there is clearly a tension between them. If to intend that X do A is, as I have suggested, to think "X shall do A," then we must distinguish between two shalls, one corresponding to "We intend . . ." and one to "I, for myself, intend. . . ." Let us represent them, respectively, as "shall_w_" and "shall_l_." I suggest that ought, as an expression of intention, is a special case of shall_w.) There are, in this case, two dimensions to the universality of moral principles as universal intentions: (1) the formal universality, or universality of application whicn can be represented by the formula, "All of us shall do A in C"; (2) the universality of the intending itself, which can be represented by modifying the above formula to read, "All of us shall_w do A in C."

Let me now bring all these considerations together. I suggest that the fact that ought-statements, unlike ordinary shall-statements, have a proper negation is built on the shared intending expressed by "ought." In other words, the syntactical intersubjectivity of ought-statements which permits Dick to contradict Tom as in D(4) above, and which consists in the existence of the form "not-ought [X doing A]" in addition to the form "ought [X not doing A]" rests on the intersubjectivity of the intention expressed by ought-statements.

Notice that I am not saying merely that the existence of shared formally universal intentions is a sociological condition of the existence of intersubjective ought-talk. I am saying that intersubjective ought-talk contains within itself the "symbolic form" which is the very existence of intending-as-one-of-us. (Just as "I shall do A," in its candid, un-self-deceived use, is the very existence of a personal intending.) For this reason, a person who said "People ought to do A in C" but denied "We intend that people do A in C" would be like a person who said "I shall do A" but denied "I intend to do A." The truth of "X intends to do A" is a necessary condition of the "genuineness" (candor and absence of self-deception) of X's "I shall do A," much as the truth of "X believes that S is P" is a necessary condition of the "genuineness" of X's "S is P." Similarly, the truth of "Group G intends that people do A in C" is a necessary condition of the "genuineness" of "People ought to do A in C" said by a member of the group. Here, however, "genuineness" is a more complicated matter than the candor and absence of self-deception of an individual. One can know that he intends that people do A in C, and yet be deceived about the group's intention. The group has shared intentions by virtue of the fact that its members intend in the mode "shall_w_." But that the members intend in this mode does not guarantee that in point of fact there are shared intentions. Intending in the mode "shall_w_" is a "form of life," a conceptual framework within which moral discourse exists and without which it is impossible. Yet the actual existence of shared universal intentions is no more an antecedent condition of participating in moral discourse than actual agreement on matters of fact is an antecedent condition of participating in factual discourse. In each case the forms of discourse set this agreement as a task. To abandon the idea that disagreement on moral matters (is even in principle capable of resolution is not to retreat to a moral solipsism; it is to abandon the moral framework itself, and to retreat to the language or "form of life" of purely personal intention.

13

We have argued that moral consciousness is a special form of _we_-consciousness, and, in effect, that one who does not intend in the _we_-mode, i.e., has no "sense of belonging to the group," cannot be said to have more than a "truncated" understanding of ought. The reader may be prepared to grant that something like a Darwinian natural selection of primitive cultures might bring it about that in the cultures which survive the internal and external dangers which beset them, people decide certain matters of conduct, ceteris paribus, on principles of the form

Shall_w_ [_if_ X is (was, will be) a person26 in C, then X (not) doing (having done) A]

where these principles are unexamined, and the fact that people acknowledge them a matter of "social inheritance." But what of examined morality? How can we combine the conception of moral action as action on principle, with the idea that the principles in question are reasonable principles? Why "Everybody ought to do A in C" rather than ". . . A' in C" or ". . . A" in C"? And if we have a reason, and if this reason is a state of affairs, does one not therefore act in order to realize this state of affairs, and no longer on principle?

Now one line of thought is that to justify a set of principles is to find that, all things considered, one wishes doings of A in C. A' in C, etc., as such to exist. But, as we saw in Section 8, this might have either of two meanings. The first is that to speak of such wishes is a misleading way of referring to the acknowledgment of these very principles themselves, in which case the "justification" amounts to the fact that we still espouse them after the factual heights and depths of the world have been explored. The bearing of "all things considered" on the continued espousal of the principles is left completely in the dark. The second is that "One wishes doings of A in C, etc., as such to exist" is intended to refer not to an approval of such doings on principle, but rather to a Iiking of these ways of doing things which, though general, is not on principle, and can be compared to other naturalistic (or, in Kant's sense, pathological) likes and dislikes, in which case the suggestion is most implausible. Is our favoring of promise-keeping as such a naturalistic one?

A more plausible line of thought is that which finds the reason for these principles rather than those in a relation they have to the general welfare. But if we have regard for the general welfare, and find it to be such a reason, does this not mean that we now act in order to promote the general welfare, and envisage doings of A in C, A' in C, etc., as means to this end? And does this not, in turn, mean that we are no longer acting on principle? Can a person have a reason for his principles, and yet, without having to forget this reason, act on these principles? Indeed, can there be such a thing as having principles and having a reason for them at all?

Now once the question is posed in these general terms, the source of the puzzle becomes clear. It lies in a failure to distinguish two senses in which a person may, at the time of acting, be said to have a certain intention. In the first place, there is the intention of the action, i.e., what he intends to do as doing that particular action. Then there are the purposes and intentions which, though he has them in mind, in an appropriate sense, at the time of acting, and though they may be closely relevant to that action, cannot correctly be said to be part of what he intends to do as doing that particular act. Plans, purposes, policies and the like do not have to be consciously entertained in order to be "in mind" and not forgotten. What I want to emphasize, however, is that even when they are consciously entertained, and as intimately related as they can be, after their own manner, to the intention of the action, they need not, for that reason, be part of the intention of the action.

I shall first develop this point in a way which oversimplifies the logical relations involved. Suppose that Jones, however he may have come to do so, loves his neighbor as his brother, and his brother as himself. "Would," he says, "that men, generally, were happy!" A study of the hearts of men and the ways of the world convinces him that an essential condition of the general welfare is that people generally do A in C, A' in C', etc. It also convinces him that for this to be the case, it is necessary that people, generally, act on the corresponding principles. (It may also convince him that general action on principle, in addition to being an indirect condition of the general welfare, is a direct condition, even, in some sense, a component, of it.)

Now it cannot be true that Jones intends that everybody act on these principles, unless it is also true that Jones intends that he himself act on these, principles.

Would that everybody acted on P1, P2, . . . Pn!
So, would that I acted on P1, P2, . . . Pn!

But his decision to acknowledge these principles is the decision to acquire (or, if already acquired, to reinforce and maintain) the character trait of arriving at decisions on matters of conduct by reasoning which, for our present purposes, can be represented by the form

Doing this would be a case of a person's doing A in C (or A' in C, etc.)
So, I shall do it

and while his efforts to acquire (and maintain) these dispositions have the purpose of doing that which is conducive to the well-being of men generally, these traits, once acquired manifest themselves in doings of A in C, A' in C, etc., of which, in an important sense, the complete intention has the form represented immediately above. Or, to put it in the language of ought,

Doing this would be a case of a person's doing A in C
So, I ought to do it.

In other words, these dispositions manifest themselves in actions of which, in an important sense, the motive is not the love of one's fellow men, but the sense of duty. f

Yet if, in an important sense, Jones does A because it is what he ought to do rather than because it will further the well-being of his fellow men, there is still a sense in which he can be said to have both intentions in doing this action. They are not, however, cooperating intentions on the same level. We must say, rather, that each particular case of conscientious action on the part of Jones is supported by his abiding intention to develop and maintain in himself the dispositions to act on the principles, i.e., to espouse them; and this intention, in turn, is part and parcel of his abiding intention that these traits of character be a common possession of men generally, which, in turn, is a consequence (given his beliefs about the hearts of men and the ways of things) of his intention that men generally he happy, a state of affairs which he wishes to exist for its own sake.

Let us take a closer took at the move from "Would that men generally were happy" to "Would that I espoused such and such principles of conduct." The logical bridge between these statements can be schematically represented (in a way which obviously oversimplifies the empirical dimension of the reasoning) as follows:

Would that men generally were happy!
(Men generally being happy) implies (men generally doing A' in C)
So, would that men generally did A' in C
(Men generally doing A' in C) implies (men generally espousing P)
So, would that men generally espoused P
(Men generally espousing P) implies (my espousing P)
So, would that I espoused P

Two objections can be raised against this reasoning. The first is that if we interpret "men generally" as "everybody," the argument founders on the fact that we know that not everybody will do A' in C and also that not everybody will espouse P. The justification of moral principles is not soft-headed sentimentality. While if we interpret "men generally" as "most people," the conclusion simply does not follow.

This objection is too strong as it stands, for it can be countered by taking the second horn and pointing out that from "Most A is B" to "(probably) this A is B" valet consequentia -- given, of course, that we have no reason to suppose that this A is one of the exceptions. Yet the objection does prepare the way for a more penetrating objection. This objection is that the empirical facts do not discriminate between "most men" and "most men with the exception of myself," and, in particular, between "most people espousing P" and "most people (with the exception of myself) espousing P." The way seems open for a compromise between benevolence and self-love which undercuts the above reasoning. One who raises this objection will grant that a person fraught with benevolence but lacking in self-love would, on sitting down in a cool hour, be in a position to reason as above. But, it is argued, surely one who combines a modicum of self-interest with even a substantial amount of benevolence would find a loophole for a compromise. That this cpmpromise would itself be the espousal of a general policy or plan of life is no answer, for it would not be the espousal of universal principles impartially applicable to all.

To rebut this objection we must rake a closer look at the concept of benevolence. For, as might be expected, the bridge between benevolence and the life of principle requires a sense of "benevolence" which logically precludes the above compromise. Is there such a sense? Once again the answer lies in reflection on the force of "we" and its relation to "I." For the sense of "benevolence" that is required is not the impartial love of everybody, which is, as Kant saw, the espousal of a principle of conduct (roughly that one ought to help other people realize their ends), but the simple recognition of people generally as one. We have already seen that moral principles involve the consciousness of us. In fact, we have found it to play a dual role in principles, as can be brought out by representing them by the form "We shall_w_ do A if in C." We can now add a third role to our list, the role which relates to the cool hour in which we rise above the level of conscientiousness as the unreflective fruit of "good upbringing." By now the direction of my argument should be obvious. For if we replace "most people but not I" by "we but not I" in the formulation of the objection, we move from consistency to incoherence. There is no logical place for a compromise between benevolence and self-love, where "benevolence" is understood as the consciousness of oneself and one's fellow men as we.

It is particularly important to distinguish the "loyalty" to people generally, the recognition of each man everywhere as one of us, from the impartial love of one's fellow men which is itself a matter of principle. For if one confuses these two, one will suspect that to defend principles in terms of impartial love is to reason in a circle. The recognition of each man everywhere as one of us was the extension of tribal loyalty which exploded it into something new. It has a precarious toehold in the world, and we are usually a far smaller group. Kant's conception of each rational being everywhere as one of us is a still more breath-taking point of view which may yet become a live option.

14

We have seen that, and how, the idea that the prime mover of reflective moral consciousness is benevolence can be reconciled with the idea that moral action is action on principle. It must be admitted that the character trait of acting on principle can exist without loyalty, as the fruit of training, precariously reinforced by praise and blame, and, on a larger scale, as a factor making for the selective survival of social groups and communities. Nevertheless, a conscience of this "chilly" kind must be threatened by every "cool hour" of critical reflection. For while self-love can find reasons for doing the things which good men do, it is unlikely to find reasons for maintaining the dispositions to act on moral principles which make good men good.

Self-love could, indeed, support the effort to acquire and maintain the disposition to arrive at a decision on certain matters of conduct by reasonings which are ultimately of a form represented by the schema

Shall [if I am in C at t, then my doing A]

that is to say, by reasoning which is built on an ego-centric intention, universal only with respect to time. Self-love could support such self-discipline on the ground that it is by being disposed to act on these principles that I shall be most likely to achieve my happiness.

But even the prudential ought is not to be confused with such egocentric rulishness. For ought, as we have seen, signalizes the presupposition of agreement. And the distinctive feature of the prudential ought is not its restriction in scope, but rather the fact that it is reflectively acknowledged that the reason for being prudentially conscientious is the fact that it is a means to one's own happiness. The prudential ought encompasses individual differences by bringing them into the content of its legislation, and, in this way, adds realism to its presupposition of intersubjectivity.

If this is correct, we should not expect the distinction between moral and prudential principles to have been made until men were on the threshold of reflective morality and had seriously begun to sit down in cool hours and raise the familiar groping questions which gave rise to moral philosophy. This does not mean, of course, that to have arrived at a reasonably clear conception of a certain class of principles as justified by self-love, and, hence, to have set these apart as "principles of prudence," one must have arrived at an equally clear conception of moral principles as justified by benevolence.

I have emphasized in the foregoing that27 the only frame of mind which can provide direct support for moral commitment is what Royce called Loyalty, and what Christians call Love of Neighbor (caritas). This is a commitment deeper than any commitment to abstract principles. It is a precious thing, the foundation for which is laid in earliest childhood, though it can arise, in adult years, by a phenomenon known, in other contexts as conversion. Recent psychological studies make clear what has always, in a sense, been known, that the ability to love others for their own sakes is as essential to a full life as the need to feel ourselves loved and appreciated for our own sakes, unconditionally, and not as something turned on or off depending on what we do. Thus, in a deeper sense, really intelligent and informed self-love supports, and can be an incentive to forming, the love of neighbor which, nevertheless, alone gives direct support to the moral point of view when we are alone in that cool hour.

APPENDIX

Since submitting this paper to the editors of Methodos I had an opportunity to read, in mimeograph, Chapter IV of Professor Castañeda's forthcoming important book on the logic of prescriptive discourse. While I disagree with certain aspects of his argument, and, in particular, with his defense of the concept of imperative inference, I do agree that the manifold of practical discourse is illuminated by viewing it against the model of a single mode of practical discourse variously enriched. Indeed, it is largely to him that I owe my possession of this insight, around which have clustered such additional insights as I have accumulated since I began my attempt to reconcile Prichardian intuitionism with naturalism some twenty years ago.

Until I read this chapter it had not occurred to me that promising and commanding ("I command . . .") admit of being viewed as specific enrichments of basic practical sentence forms. Thus, while I would have explained the fact that there are no such reasonings as

(1)

I promise to do A, if p

p

So, I promise to do A

or

(2)

I command you to do A, if p

p

So, I command you to do A

as opposed to

(1a)

I promised to do A, if p

p

So, I promised to do A

and

(2a)

I commanded you to do A, if p

p

So, I commanded you to do A

by saying that the signals "I promise" and "I command" do not belong in the scope of "---, so . . . ," this claim, sound in essence though it is, would have been scarcely more than the report of an isolated "logical intuition."

In the terminology of this paper, "I promise to do A, if p" differs from "I shall do A, if p" (and, similarly, ''I command you to do A, if p" from "You shall do A, if p!") in that they are not only tellings but tellings which, by virtue of the presence of the phrases "I promise" and "I command" are (if appropriately performed) subsumed under specific moral principles, as explained in Section 2. "I promise to do A, if p" has, therefore, the form "I shall do A, if p*," and "I command you to do A, if p" has the form "You shall do A, if p**," where "*" and "**" represent, respectively, the promise-making and the command-issuing signals. Like the telling-to signal "!'", these signals are not, as such, ingredients in practical reasoning. But if there are no such practical reasonings as (A) and (B) above, there is the reasoning

First step:

p

So, (its being the case that if p, then X does A) implies (X doing A)

Second step:

So, (Shall [its being the case that if p, then X does A]) implies (Shall [X doing A])

Third step:

Shall (its being the case that if p, then X does A)

So, Shall (X doing A).

I also wish to take this opportunity to thank Dr. Karl Potter, whose helpful criticism of the previous version of this paper led me to substitute the convention according to which "X shall do A" is the sheer expression of an intention, whereas "X shall do A!" is a telling to as well, for a considerably less perspicuous representation of this distinction.

NOTES

1. This paper is a revised version of a paper which appeared in Methodos, VIII (1956), 227-268.

2. W. D. Ross, The Right and the Good (Oxford: Clarendon Press, 1930).

3. For an interesting discussion of a parallel point, see Professor C. D. Broad's animadversions on the concept of a "rational being" in "Some Reflections on Moral Sense Theories in Ethics," Proceedings of the Aristotelian Society, Vol. XLV (1944-45), pp. 131-166.

4. "Does Moral Philosophy Rest on a Mistake?" (1912), in Readings in Ethical Theory, ed. W. S. Sellars and J. Hospers (New York: Appleton-Century-Crofts, 1952), p. 154.

5. Ibid., p. 155.

6. A. J. Ayer, Language, Truth and Logic (London: Victor Gollancz, 1936).

7. The Language of Morals (Oxford: Clarendon Press, 1952). My understanding (to the extent that it is not misunderstanding) of Hare's view has been enhanced by an exchange of ideas in correspondence and personal conversation for which I am gratetul.

8. May I take this occasion to acknowledge my indebtedness to my friend and former student, Dr. Hector-Neri Castañeda, whose dissertations for the M.A. and Ph.D. degrees from the University of Minnesota contain important and original contributions to the imperativists' approach to practical discourse. I have also had the benefit of several lengthy discussions with him in the early spring of 1956. The present paper contains many traces of his influence, though I do not know how much of its detail he would find acceptable.

9. I am concerned in this section with the "implicit universality" of the moral ought. I do not wish, therefore, to be taken to deny that there might be varieties of ought which have a lesser scope than all persons at all times. An ought, however, which simply reflected one person's intentions here and now would be no ought at all.

Even the sense in which the moral ought applies to all persons requires careful analysis. It should not be assumed that the "everybody" in a moral principle has the force of "all human beings" in a purely descriptive, so to speak biological, sense of this phrase. It means, roughly, all of us, where we are those who accept each other as relevant participants in a discussion of what ought or ought not to be done, rather than simply belonging to the stage-setting of conduct. Today, we are men generally.

10. "Presupposing," Philosophical Review, LXII (1953), p. 214, n. 9.

11. It is, perhaps, worth noting that singular cause-statements are also, in this same sense, "implicitly universal." Thus a person who says, "Doing this to that in these circumstances caused it to behave thusly,'' commits himself to support his remark with a statement of the form, "This is an X, and doing that to an X is Y-ing it, and these circumstances are of kind C, and behaving thusly is Z-ing, and Y-ing an X in C causes it to Z. Singular cause-statements, then, like singular ought-statements, are not merely singular. Both are, so to speak, singular "after their own kind." Notice also that one can claim with reason that this caused that in these circumstances even though one does not have a causal universal "up one's sleeve" to cover one's claim. One can have reason to believe that the circumstances are such, and this and that of such kinds, that "there is" a causal universal under which they could be subsumed.

12. John Austin, "Other Minds," Proceedings of the Aristotelian Society, Suppl. Vol. XX (1946), pp. 169-174.

13. For an explication of the concept of defeasibility, see H. L. A. Hart, "The Ascription of Rights and Duties," Proceedings of the Aristotelian Society, Vol. XLIX (1948-49), pp. 171-194.

14. Op. cit., pp. 24ft.

15. Since we are concerned with modal contexts, it is important not to suppose that if P strictly implies Q, then (Shall [P]) implies (Shall [Q]). Stronger requirements are necessary to avoid paradox. As far as I can see, something like A. R. Anderson's reconstruction of "entails" is necessary.

16. It will be noticed that I am referring to statements of intention as statements. This seems to be in accordance with ordinary usage, although logicians have tended to restrict the term to items which are either true or false.

17. Hypothetical imperatives, represented by the form "If you want A, do B!" are, of course, not to be confused with the conditional imperatives we were discussing a moment ago. The latter have the form "Do B, if p!" where p is any relevant statement of fact. Hare has made it clear that "if you want A" is not a descriptive-psychological condition serving as a special case of p, but rather specifies a resolutive premise which is thought to be (possibly) operating in the practical reasoning of the person to whom the hypothetical imperative is addressed.

18. Hare, of course, construes "imperative" so inclusively that it covers "I shall do A" as well as "Do A!" Yet, since he construes all imperatives as tellings to, he construes "I shall do A" as telling oneself to do A, which it clearly is not; though, of course, there is such a thing as "telling oneself to do A."

19. There are, of course, differences. It makes sense to say of a person that he intended to do or bring about something which he did not want to do or bring about. If one does B in spite of the fact that it brings about A, then one does not want to bring about A, although one intends to do so. It should also be clear that one may intend and, indeed, want that X do A although one does not desire that X do A. We do not use the word "desire" (or its cognates) in the case of action on principle.

20. For an earlier statement of this point, see my "Obligation and Motivation," Philosophical Studies, II (1951), 21-25. In the Introduction to the second edition of his Language, Truth and Logic, Ayer criticizes his earlier formulation of the emotive theory as follows: ". . . I fail to bring out the point thar the common objects of moral approval or disapproval are not particular actions so much as classes of actions: by which I mean that if an action is labeled right or wrong, or good or bad, as the case may be, it is because it is thought to be an action of a certain type." Ayer is here on the point of recognizing the existence of practical reasoning as a genuine form of reasoning. But by failing to draw the above distinctions, he remains within the limits of his earlier formulation.

21. If we suppose that this principle can be put in the form

If anybody were in C, would that he did A!

we notice that if this is confused with

Would it were the case that (if anybody were in C, he did A)!

one arrives at the Kantian formulation.

Would that it were a law of nature that people in C did A!

22. For an elaboration of such a framework, see my "Empiricism and the Philosophy of Mind," in Minnesota Studies in the Philosophy of Science, Vol. I (Minneapolis: University of Minnesota Press, 1956), pp. 253-328. See also "Some Reflections on Language Games," Philosophy of Science, XXI (1954), 204-228.

23. To appreciate at least the general force of this point, one has only to reflect that one has not learned to use "shall" unless, ceteris partbus, candid utterances of "I shall do A" are followed by the actual doing of A. "I shall do A" is, so to speak, "My doing of A in the future" as signalizing a forthcoming doing of A. Furthermore, the forthcoming doing of A which it signalizes is not a blind doing of A, and this in two respects: (1) the doing is, in a sense, brought about by the idea of doing A; (2) the bringing about itself is a self-conscious fact, and not a mere triggering of the action by the idea. This self-consciousness consists in the fact that the thought responsible for the action is "I shall do A," and can be traced to the fact that one hasn't learned the use of "shall" unless "I shall do A" is not only the signal ol a forthcoming doing of A, but is understood to be such a signal.

These and other points concerning the interrelationships of the order of reasons and the order of causes will be developed in a separate paper which began as a section of an early draft of the present paper.

24. I am assuming, of course, that the possibility of X doing A has occurred to Tom, and that he has been deliberating about it.

25. I am coining this expression because "countermanding" as ordinarily used appears to cover both simple withdrawal and what I am calling "contramanding."

26. See note 9 for an interpretation of "person" as "one of us."

27. These concluding remarks were taken from an unpublished paper, "Ethics and Philosophy," which was read to the Phoebe Griffin Noyes Library Association, Old Lyme, Connecticut, on January 26, 1960.