Form and Content in Ethical Theory

Wilfrid Sellars


The Lindley Lecture for 1967 (Department of Philosophy, University of Kansas, 1967).
Transcribed into hypertext by Andrew Chrucky, Sept. 13, 1997.

1. The focal point of practical reasoning is action, as the focal point of empirical reasoning is observation. Perceptual takings or 'judgments' are the thoughts which typically arise from the impact of the world on our mind through our sensory capacities. Volitions are the thoughts which typically impinge on the world through our motor capacities.

2. Intentions can be thought of, somewhat metaphorically, as practical commitments. Volitions can correspondingly be thought of as practical commitments to do something here and now, and hence as a special case of commitments to do something at sometime or other. If one comes to intend to do a certain action in the future, say ten minutes from now, whether impulsively or as a result of deliberation and decision, then, unless one changes one's mind, the intention becomes a here-now intention when one realizes the ten minutes have gone by. More generally, if one has the intention to do a certain action if a certain circumstance should obtain, then, unless one changes one's mind, the intention becomes an unconditional intention when the circumstance is believed to obtain.

3. I shall assume that intentions uniformly find their linguistic expressions in the use of the auxiliary verb 'shall.' The intention to do A in ten minutes would be expressed by
I shall do A in ten minutes.
The intention to do A when and if C obtains would be expressed by
I shall do A when and if C.1

4. Intentions are not limited to intentions to do, whether now, or later, or on the condition that a certain circumstance obtains. There are also intentions that something be the case. The latter, however, are intentions, practical commitments, only by virtue of their conceptual tie with intentions to do. Roughly
It shall be the case that-p
has the sense, when made explicit, of

(Ceteris paribus) I shall do that which is necessary to make it the case that-p.

5. It is important to see that I can not only intend to do something myself, I can also intend that someone else do something, i.e. that it be the case that he does it. Intentions pertaining to the actions of others are not 'intentions to do' in the primary sense in which
I shall do A
is an intention to do. Thus, in spite of their superficial similarity,
Tom shall do A
and
I shall do A
do not have the same conceptual structure. The former has the form

(Ceteris paribus) I shall do that which is necessary to a make it the case that Tom does A

whereas the latter cannot, without the absurdity of an infinite regress, be supposed to have the form

(Ceteris paribus) I shall do that which is necessary to make it the case that I do A.2

6. These considerations highlight the fact that the intention expressed by a 'shall' statement is invariably the speaker's intention. Thus
Tom shall do A
expresses the speaker's intention that Tom do A. This 'first person' feature of intentions consists in part in their relation to the
I shall do
which can become the commitment to do something here and now which is volition.

II

7. Intentions imply intentions just as beliefs imply beliefs. This point must be carefully made. We must distinguish between intentions as states of intending and intentions as what is intended, just as we distinguish between states of believing and what is believed, the so-called 'content' of the believing. In the latter case we distinguish between the implications of the content of a belief and the implications of the state of having a belief with that content. Thus the belief which would be expressed by the conjunction of Peano's postulates (P) implies the belief which would be expressed by any arithmetical theorem (T), however recondite, in the sense that the one belief content implies the other. Yet, obviously, the existence of this implication does not carry with it the idea that
Jones believes P implies Jones believes T.

8. Corresponding distinctions obtain in the case of intentions. Thus, when I speak, as I shall, of one intention I1 implying another intention I2, I shall be speaking about an implication between two intention-contents. That in this sense an intention I1 implies an intention I2 does not carry with it the idea that
Jones intends I1 implies Jones intends I2.
An ideally rational being would intend the implications of his intentions, just as he would believe the implications of his beliefs.3

9. Philosophers analyse the logical relations of belief-contents by determining the logical relations of the factual statements which express them. In this lecture I shall explore the logical relations of intention-contents by exploring the implications of the practical statements which express them.

III

10. Philosophically, the most important single feature of the logic of intentions is that it is parasitical upon the logic of beliefs. This fact can be expressed as the principle
'It shall be the case that P' implies 'It shall be the case that Q' <--> 'P' implies 'Q.'
The technical elaboration of this principle would transcend the scope of this lecture. I have argued elsewhere4 that, for example, the reasoning
I shall I do A, if p
p
So, I shall do A
can be transformed to fit the following principle of inference:
'Shall [if p, my doing A]' implies (relatively to the assumption that p) 'Shall [my doing A]'.

11. If 'P' implies 'Q,' then it is unreasonable to believe that P is the case without believing that Q is the case. (Though, as noted above, in point of fact one may well believe the former without believing the latter). Similarly if
'It shall be the case that P' implies 'It shall be the case that Q'
it is unreasonable to intend that P be the case without intending that Q be the case. (Though, again, in point of fact one may very well intend the former without intending the latter and may even intend that the latter not be the case.)

IV

12. The central theme of Kant's ethical theory is, in our terminology the reasonableness of intentions. In what sense or senses, if any, can intentions be said to be reasonable, i.e. have a claim on the assent of a rational being? Kant clearly construes this task as parallel to the task of defining in what sense or senses, if any, beliefs can be said to be reasonable, i.e. have a claim on the assent of a rational being. As in his epistemology, Kant sides with the rationalists against both the empiricist and the skeptic -- but gives rationalism that twist which makes all the difference. In both areas his insights were so revolutionary that they are even now just beginning to be absorbed.

13. The primary distinction Kant draws with respect to the reasonableness of intentions, is that between 'hypothetical' (or, as I prefer to put it, 'relative') and 'categorical' reasonableness. The simplest examples of intentions whose reasonableness is purely 'relative' are provided by what Kant calls 'hypothetical imperatives'.

14. A hypothetical imperative (the term 'imperative' is, of course, a misnomer) has the form
If S wants to bring about E, he ought to do A
or, more explicitly,
(Since S is in C)
If S wants to bring about E, he ought to do A
since the assertion of a hypothetical imperative in the unqualified form presupposes that the person in question is manifestly in a circumstance which makes the imperative relevant.

15. What does the hypothetical imperative assert? The distinctions we have drawn enable an answer which is free from the puzzles which have traditionally hovered around this question. It asserts that (since S is in C) the intention-content pertaining to his bringing about E implies the intention-content pertaining to his doing A. In other words
'I, S, shall bring about E' implies 'I, S, shall do A'
or equivalently
'I shall bring about E' implies (for S) 'I shall do A'.

16. Notice that it is important to use the personal pronoun 'I,' for the implied intention is a 'shall do'. Furthermore, since all of us are 'I's, but only S is presumed to be in C, we write 'implies (for S),' for the implication hinges on S's being in C. (For a deeper analysis of the reason for relativising the implication see V-27 below.)

17. In the terminology suggested above, the hypothetical imperative asserts that (since S is in C) the intention which he would express by saying
I shall do A
is reasonable relative to the intention which he would express by saying
I shall bring about E.

18. One merit of using the term 'relative' to characterize the reasonableness of the above intention, is that it enables us to avoid confusing the reasonableness (for S) of the intention to do A relative to the intention to bring about E, with a supposed reasonableness (for S) of the intention to do A on the hypothesis that he intends to bring about E.5 This confusion generates the mistaken idea that the hypothetical imperative authorizes S (since he is in C) to reason
I intend to bring about E
So, I shall do A6
whereas the hypothetical imperative actually says that (given that S is in C) a reasoning on his part to the effect that
I shall bring about E
So, I shall do A
would be valid, in that the premise does (given that S is in C) imply the conclusion.

19. Thus the reasonableness involved by a hypothetical imperative is the reasonableness of a conclusion intention relative to the premise intention in a (possible) piece of practical reasoning. It does not commit itself concerning the reasonableness of either the premise intention or the conclusion intention per se.

20. Let me prepare the way for the next stage of the argument by reminding you of the distinction between the validity and the goodness of arguments in the domain of theoretical reasoning. An argument can be valid, but fail to be good, by having a false premise. To say that it is valid, is, in our terminology, to say that its conclusion is reasonable relative to its premise. To say that it is good is to add that its premise is reasonable -- for, I shall assume, truth is a special case of reasonableness. A true proposition is one which has a certain claim to be assented to by a rational being. Just how this claim is to be analyzed is the problem of truth which, fortunately, must be left to other occasions.

21. One would expect, then, that a good practical argument is one in which the conclusion intention is not only implied by the premise intention, so that it is reasonable relative to the premise, as in the case of arguments authorized by hypothetical imperatives, but also one in which the premise intention is reasonable per se. And, indeed, Kant is clearly looking for a property of intentions which corresponds to truth. In short, he is attempting to discover what might make practical arguments good as opposed to merely valid. My purpose in this lecture is to show that he took us to the very threshold of success.

V

22. But before we make a frontal assault on the problem of what might make an intention categorically reasonable, or, to use Kant's terminology, valid, we must clarify and broaden the concept of the relative reasonableness of an intention.

23. Hypothetical imperatives typically rest on causal connections, and, like most singular causal statements in everyday life, they are rarely if ever the direct application of a general causal law. Thus
If Jones wants a drink, he ought to go to the next corner
is not the application of a supposed general law to the effect that
If anyone is to get a drink he must go to the next corner.

24. We have already noted that hypothetical imperatives typically presuppose that the person in question is in a certain circumstance. This point must now be elaborated. The first step is to bring this presupposition into the content of the imperatives, thus
(Since S is in C)
If S wants to bring about E, he ought to do A
becomes
If S wants to bring about E, he ought to do A, if he is in C and S is in C.
Let us focus our attention on the complex if statement, and neglect the conjoined assertion.

25. There are many ways in which general lawlike statements can be idealized. Thus it is often required that a 'genuine' lawlike statement contain no reference to particular objects, times or places. Whether or not ideal science would give us such, it is clear that we often have to settle for less. Now the nomologicals with which we are concerned are those which can generate general hypothetical imperatives. These nomologicals concern the causally necessary conditions for bringing about a certain kind of state of affairs in a certain kind of circumstance.7 They have the form
Doing Ai if Cj is causally necessary to realize to the realization of Ek.
Or, putting it in terms of causal implication, we have the family of implications,
'x brings about Ek' implies 'x does Ai if in Cj'.
These implications, which for obvious reasons can be called 'instrumental implications' are 'binding on all rational beings' in the sense that as empirical generalizations, their inductive soundness is independent of the desires and inclinations or cultural ties of specific individuals or groups. Transposed into practical discourse as a general hypothetical imperative, they become
'I shall bring about Ek' implies 'I shall do Ai, if in Cj'.
Even after this transposition, the implications remain binding on all rational beings. Any restriction belongs in the circumstance clause. Thus to restrict it to WASPs is to include the characteristic of being a WASP in Cj. For if being a WASP is irrelevant to bringing about Ek by doing Ai in Cj, then there is no point in including it anywhere; while if it is relevant, the relevance is a causal one and belongs in the content of the implication and not as a limitation on those for whom it holds.

26. Thus a limitation of the general hypothetical imperative to WASPs will not take the form
'I shall bring about Ek' implies (for all WASPs) 'I shall do Ai in Cj'
but
'I shall bring about Ek' implies 'I shall do A, if in CjM
where 'CjM' differs from 'Cj' by including the additional characteristic of being a WASP. In this sense, general hypothetical imperatives can be said to hold "for all rational beings." They are simply the transposition into practical discourse of empirical instrumental generaIizations.

27. Notice, however, that although general hypothetical imperatives "hold for all rational beings", there is an important sense in which each such imperative formulates not one single implication but a family of implications, one for each rational being. This complication reflects the fact that 'I' is a systematically ambiguous term. Thus, general hypothetical imperatives have the form
'I shall bring about Ek' implies (for each rational being) 'I shall do Ai, if in Cj.
A general hypothetical imperative asserts that each rational being can validly argue
I (Tom) shall bring about Ek
So I (Tom) shall do Ai if in Cj
I (Dick) shall bring about Ek
So I (Dick) shall do Ai if in Cj

etc.

VI

28. Let us now make a preliminary attempt to understand what it might mean to say of an intention that it is 'categorically reasonable' or 'categorically valid.'

29. Let us suppose that in the circumstance in which I now find myself I ought to do a certain kind of action A. The following categorical ought statement, we shall suppose is true.
I ought to do A.
Standing behind this categorical statement, however, is a condition of the form 'since (or because) I am in C,' we are thus led to the conditional ought statement
If I am in C, I ought to do A.
This in turn points in the direction of
Anybody ought to do A, if he is in C
but before attempting to understand the logical status of the reference to anybody, let us dwell for a moment on the first person conditional.

30. The first point to be made is that the antecedent of this conditional refers to a circumstance rather than, as in the case of
If I want to bring about E, I ought to do A
to an intention. Thus if the ought statement tells us that the intention to do A is reasonable, it does not tell us, at least explicitly, that it is reasonable relative to another intention. Its explicit message is that the intention to do A is reasonable relative to the condition of being in C. On the other hand, if we accept, as we have, the principle that intentions can only be derived from other intentions, this reasonableness points to the argument
I shall do A, if I am in C
I am in C
So, I shall do A
in other words, implicitly the reasonableness of 'I shall do A' is relative not only to the circumstance, but to the conditional intention
I shall do A, if I am in C.
It is, therefore, this intention which must be categorically reasonable, if the original ought statement is to express a categorical imperative.

31. We are thus led to the idea that
I ought to do A, if I am in C
is equivalent to
'I shall do A, if I am in C' is categorically reasonable.
Note that in this statement categorical reasonableness is predicated of a conditional intention. 'Categorical' in the sense which applies to reasonableness must not be confused with 'categorical' as a classification of propositional forms.

32. But what are we to make of the idea that an intention of the form
I shall do A, if I am in C
can be categorically reasonable? Obviously the reasonableness does not consist in the logical form of the intention. Intentions of this form do not show themselves to be sound, for example by unpacking into tautological intentions such as
I shall either stay or go, if I am threatened,
or unsound by unpacking into contradictory intentions such as
I shall stay and go, if I am threatened.

33. But how can the specific subject matter of an intention be involved in its reasonableness, without turning the latter into a disguised form of the relative reasonableness asserted by a hypothetical imperative? Curiously enough, the key to the answer is found by seeing how close we can come to capturing the distinctively categorical reasonableness of morally sound intentions by construing it as a special case of the relative reasonableness ascribed to intentions by hypothetical imperatives.

34. Is there a hypothetical imperative in the neigborhood which has a moral ring to it? The obvious candidate is
If S wants to maximize the general welfare (GW), S ought to do Ai if S is in Ci.
Notice that this hypothetical imperative combines the two modes of conditionality, expressed, respectively, by 'if S wants' and 'if S is in C'. According to our analysis, this hypothetical imperative tells us that
'I shall maximize GW' implies (for S) 'I shall do A if I am in C'
The implication is, as we have seen, one of our special kind we have called 'instrumental.' And if the implication holds for S, it holds for any rational being, for we are supposing, as before, that the hypothetical imperative is simply the transposition into practical discourse of an inductively established empirical generalization.

35. This complex hypothetical imperative as such asserts the reasonableness of the conditional intention
I shall do A, if I am in C
relative to the intention
I shall maximize GW.
We can now point out that if the antecedent intention was itself categorically reasonable, and if, as we have been assuming, categorical reasonableness is the practical counterpart of truth, it would follow that the consequent intention was itself categorically reasonable.

36. It has been easy to assume that relative and categorical reasonableness are incompatible; that an intention can have one or the other, but not both.8 This assumption is simply false.

37. To bring out the implications of this point, remember that a good theoretical argument is one in which (a) the conclusion is reasonable relative to the premse, and (b) the premise itself is categorically reasonable, i.e. true. Implication preserves truth in theoretical arguments. We should explore the possibility that it preserves categorical reasonableness in practical arguments. If so, then, an intention can be categorically reasonable, and yet derivative from another intention -- provided, of course, that the latter in turn is categorically reasonable.

38. Categorical reasonableness must not be confused with intrinsic reasonableness. The confusion between these two has been even more damaging to Kant exegesis, than the tendency to suppose that a categorically reasonable intention cannot be conditional in its logical form. On the other hand even if categorical reasonableness is not the same as intrinsic reasonableness, we are faced with the fact that if there are to be derivative categorically reasonable intentions, there must be one or more intentions whose categorically reasonableness is non-derivative or intrinsic. Are any to be found?

39. What of the antecedent of the above complex hypothetical imperative
I shall maximize the GW?
It is a worthy intention, one that we should encourage people to have -- though not, as Kant emphasizes, at the expense of the sense of duty. Yet it does not seem to have any feature which calls for the predicate 'intrinsically and categorically reasonable.'

VII

40. Let us continue to beat about in the neighboring bushes. The first thing to note is that the instrumental nomologicals on which the above complex hypothetical imperatives rest generate not only hypothetical imperatives, but other practical implications which, though closely related, are not in the strict Kantian sense hypothetical imperatives. For not all statements to the effect that one intention implies another can be put in the form of a hypothetical imperative, at least if we tie this term to Kant's paradigms.

41. To develop this point, we must remember that not all intentions are intentions to do. There are also intentions that something be the case. And if the existence of one state of affairs causally implies the existence of another, then the intention that the one obtain causally implies the intention that the other obtain.

42. Thus the instrumental nomologicals which generate the general hypothetical imperatives
'I shall maximize GW' implies (for each rational being) 'I shall do Ai, if I am in Ci'
also generates, for example,
'It shall be the case that Tom (Dick, Harry) maximize GW' implies (for each rational being) 'It shall be the case that Tom (Dick, Harry) does Ai, if he is in Ci'
and, indeed, the doubly general practical implication
For all values of 'x,' 'it shall be the case that x maximizes GW' implies (for each rational being) 'it shall be the case that x does Ai if x is in Cj'.9

43. Here the 'shall's are 'shall be the case's and though the egocentric 'I' has apparently dropped out, it is still present by virtue of the conceptual relationships between 'it shall be the case . . .' and 'I shall do . . .'

44. The difference between the general hypothetical imperatives and the general practical implications schematized in 42 lies in the fact that whereas the former authorize each person to reason about himself, thus
I shall maximize GW
So, I shall do Ai if in Ci
the latter authorize each person to reason about anybody, including, of course, himself. It authorizes Tom to reason about himself, thus
It shall be the case that Tom maximizes GW
So it shall be the case that Tom does Ai, if in Ci
and also about Dick, thus
It shall be the case that Dick maximize GW
So, it shall be the case that Dick does Ai, if in Ci.

VIII

45. We are now ready for the thirty-two dollar question. We have been grooming categorical reasonableness to be the practical counterpart of truth. But in theoretical reasoning truth, and hence the goodness of arguments, is intersubjective.

46. Consider the argument offered by Tom,
There is lightning at tp
So, there was thunder at t + delta-t so, ~q
This is, we shall assume, not only valid given the familiar law of nature but good, i.e. it is true that there was lightning at t.

47. If so, then Dick's argument
There was no thunder at t + delta-t~q
So, there was no lightning at tso, ~p
though equally valid, can't also be good. This is because Tom and Dick are contradicting one another, when Tom says "p" and Dick says "~p".

48. Tom's practical reasoning,
It shall be the case that Dick maximizes GW
So, it shall be the case that Dick does Ai if in Ci
although it is valid, as being in accordance with an implication which is binding on each rational being, is essentially private. In spite of the fact that Tom is reasoning validly about Dick, and that he would be reasoning with equal validity if he reasoned in the same way about anybody, including himself, his argument does not have the intersubjective status which would make possible a logical clash of his argument with Dick's equally valid argument
It shall not be the case that Dick does Ai if in Ci
So, it shall not be the case that Dick maximizes GW.
Dick's
It shall not be the case that Dick maximizes GW
does not stand to Tom's
It shall be the case that Dick maximizes GW
as Dick's
It is not the case that there was lightning at t
stands to Tom's
It is the case that there was lightning at t.

49. Two people can affirm the same proposition in a strong sense of 'same'. But as far as the intentions we have so far considered are concerned, intentions can at best be parallel. They are irreducibly egocentric, even when this egocentricity is latent as in
Tom: it shall be the case that the war ends
Dick: it shall be the case that the war ends.
This dialogue provides an excellent example of "agreement in attitude." But if the depth form of these statements is
Tom: (Ceteris paribus) I (Tom) shall do what I can to end the war
Dick: (Ceteris paribus) I (Dick) shall do what I can to end the war
the agreement in attitude is not an identity of intention.10

50. What of
Tom: we shall do what we can to end the war Dick: we shall do what we can to end the war.
These statements in the first person plural have the interesting properties that (a) they express the speaker's intention, yet (b) the intentions expressed are in the strongest sense the same. Put in terms of the distinctions I drew in my opening remarks, the intendings are two in number, but the content of these intendings is the same, in as strong a sense as the content of the two believings expressed by
Tom: There was lightning at t
Dick: There was lightning at t
is the same. I shall put this by saying that the intendings expressed by
We shall do . . .
have an intersubjective form.

51. This intersubjective form stands out when it is a matter of intendings to do. When, however it is a matter of intendings that something be the case: the distinction is likely to be lost, unless we index the 'shall' to indicate the form of the intending to do which the intendings imply.11 Thus the intersubjective intention expressed by
It shallWE be the case that the war end
would contrast in form, for example, with the personal intention
It shallI be the case that the roses be planted.

IX

52. We have answered the thirty-two dollar question by finding the necessary dimension of intersubjectivity. There remains, however, the sixty-four dollar question of categorical reasonableness. Some paragraphs back, when we were beating about in the neighboring bushes, I asked if the antecedent of a certain complex hypothetical imperative could be construed as categorically reasonable or valid, pointing out that if so, its categorical reasonableness would be transmitted to that which is instrumentally implied. Since that time we have (a) taken into account intentions that something be the case, and (b) brought into the picture intentions which have intersubjective form. Have these additional resources brought us closer to our goal? I believe they have.

53. Consider the intersubjective intention
It shallWE be the case that our welfare is maximized.
If this intention were intrinsically categorically reasonable, or valid, then, by virtue of the relation of 'shall be the case' to 'shall do,' so also would
We shall each of us so act as to maximize our welfare.

54. We now take into account the vast number of complicated instrumental nomologicals which we suppose to have been established (in principle) by the behaviorial sciences, and are represented by the schema
'x so acts as to maximize GW' implies 'x does Ai if in Ci'.
Given these implications the categorical validity of the above intention entails the categorical validity of the family of intersubjective intentions
It shallWE be the case that each of us does Aiin Ci
which on our analysis, is equivalent to the family of ought statements
If any of us is in Ci he ought to do Ai.

55. Notice that although the validity which these ought statements ascribe to the intersubjective intentions with respect to any of us that he do Ai in Ci is a derivative validity, it is nevertheless a categorical validity. The 'ought' itself can properly be characterized as categorical, and contrasted with the 'relative' ought of the hypothetical imperative.

X

56. If the strategy sketched in the preceding paragraph were successful, we would have succeeded in vindicating Kant's analysis of the nature and validity of moral principles. But the strategy was only sketched, and any attempts to elaborate it must deal with two serious questions.

57. The first is the most difficult. Who is 'us'? Kant believed himself to have established that 'we' must be "rational beings generally." At first sight, however, this seems to be a mistake, a confusion between the class of those for whom instrumental implications are binding, which does, indeed, consist of rational beings generally, and the class of those for whom categorically valid intersubjective intentions are binding.

58. The distinction involved can best be made by returning to the general hypothetical imperatives of benevolence:
'I shall maximize the GW' implies 'I shall do Ai, if in Ci'.
We pointed out that the scope of the 'I' consists of rational beings generally, since the implications are simply the transposition into practical discourse of inductively established nomologicals. We failed, however, to emphasize that the welfare to be maximized is the welfare of a group to which a particular 'I' belongs, thus
'I shall maximize the GW of my group' implies (for each rational being) 'I shall do Ai if in Ci'.
And if we ask which group is that, we see that it by no means follows from the above that the answer must be 'rational beings generally.' We must draw at least a conceptual distinction between the class of those on whom the implications are binding, and the class of those the promotion of whose welfare is the object of the benevolent intention of a particular 'I'.

59. The distinction remains when we transpose the instrumental implication into the realm of intersubjective intentions.
'It shallWE be the case that each of us so acts as to maximize our welfare' implies 'It shallWE be the case that each of us does Ai if in Ci'.
As before, the 'implies' can be glossed with 'for each rational being.' But, as before, it by no means follows that the group whose welfare is 'our' welfare consists of rational beings generally.

60. It might, however, be argued that only if the 'we' of 'our welfare' is the 'we' of 'we rational beings generally,' is an intersubjective intention of this form categorically valid. This might, as we shall see, be true if the welfare in question is what might be called epistemic welfare, but not if we take into account, as we must, needs and desires generally.

61. Now the categorical validity of an intersubjective intention of the form
It shallWE be the case that our welfare is maximized
would seem to consist in the fact that it is by virtue of such an intention that a group or community is a group or community. Roughly a community consists of indi\iduals who intend sub specie such an intention, the scope of 'we' being the members of the community.

62. This is not to say, of course, that there will be agreement as to just what is instrumentally implied by this intention, or that on particular occasions the implications a person believes it to have will prevail against an alternative arrived at by practical reasoning "from a personal point of view."

63. Can we say that rational beings generally constitute a community? They would do so if they shared the intersubjective intention
It shallWE be the case that each of us rational beings so acts as to promote our welfare.

64. Now, since an individual can have an intention of intersubjective form even if no one else in point of fact shares it, an individual rational being could have an intention of the above form even though few if any other rational beings had such an intention. To have this intention is to think of oneself as a member of a community consisting of all rational beings. It is possible, therefore, for a rational being to think of himself as a member of such community, even though this community does not actually exist.

65. If, however, the following two premises were established, this community could be shown to be a reality:
(a) To think of oneself as a rational being is (implicitly) to think of oneself as subject to epistemic oughts binding on rational beings generally.
(b) The intersubjective intention to promote epistemic welfare implies the intersubjective intention to promote welfare sans phrase.
These premises would entail that the concept of oneself as a rational being implies the concept of oneself as a member of an ethical community consisting of all rational beings. To be sure this implications need not be recognized. Indeed, it would take all the dialectical skill of a Socrates, a Hegel or a Peirce to bring it to the surface. Yet if the above premises were true, all rational beings should 'implicitly' think of themselves as members of an ethical community consisting of all rational beings. But since a community exists if the relevant individuals think of themselves as its members, the ethical community of rational beings would have an 'implicit' existence.

66. The first of the above premises is not implausible. If we accept it we can conclude that 'implicitly' all rational beings constitute an epistemic community. The second premise, despite Peirce's valiant efforts, remains problematic, and without it the argument for the reality of an ethical community consisting of all rational beings, the major premise of which is the "fact of reason," remains incomplete.

XI

67. I shall conclude by drawing the implications of the above analysis for certain traditional puzzles pertaining to Kant's ethical theory. For this purpose the most significant feature of this analysis is the point that the categorical validity of an intention can be derivative.

68. It is this fact which enables us to see how 'teleological' and 'deontological' themes are harmonized in Kant's ethics. Thus specific moral principles are categorical oughts, but the categorical validity of the intersubjective intentions that any rational being in a certain kind of circumstance do a certain kind of action12 is derivative from the categorical validity of the intersubjective intentions that our welfare be maximized. Thus when Kant speaks in the Metaphysical Elements of Ethics of the happiness of others as a categorical end,13 what he says is in no way inconsistent with his claim that the ought of moral principles is categorical rather than the hypothetical ought which pertains to the relation of means to ends.

69. Again, when Kant stresses intentions, he is not disregarding consequences. It is because doing Ai in Ci maximizes the general welfare that the intention to do Ai in Ci is categorically valid. Of course we may be, and often are, mistaken about what kind of action in what kind of circumstance will promote the general welfare, but what we ought to do hinges on what would actually happen. On the other hand, the moral character of our motive is a function of what we think will happen as a result of our action, though not of this alone.

70. When Kant insists that we ought to act from a sense of duty he is not making the absurd mistakes which have often been attributed to him.14 He is simply repeating the point with which he opens the argument of the Fundamental Principles of Metaphysic of Morals, that the only unconditional good is a good will. By this he means that the only state of a person which is unconditionally good from a moral point of view is the disposition to act from a sense of duty. He has two points in mind: (a) Whereas action from any motive can have bad results, the sense of duty alone is such that only by virtue of ignorance does it have bad results. Action from other motives even where ignorance is absent can lead to bad results. Thus the sense of duty is the only motive which has a direct conceptual tie to the categorically valid end of moral conduct. In this sense a good will is a categorical ought-to-be. (b) Although the general welfare is also an end in itself, a categorical ought-to-be, the ought-to-be of the happiness of any given individual is, Kant believes, conditional on his having a good will.

71. As Broad has pointed out, Kant is not always clear about the respective status of specific categorical imperatives (categorically valid maxims) and higher order principles about what distinguishes categorically valid maxims from those which are not. Thus when he writes in The Critique of Practical Reason
The principle of happiness may, indeed, furnish maxims, but never such as would be competent to be laws of the will even if universal happiness were made the object. For since the knowedge of this rests on mere empirical data . . . it can supply only general rules, not universal15
he confuses the sound point that the intersubjective validity of the intention to maximize universal happiness cannot be explicated in terms of benevolence, with the unsound idea that empirical data are not relevant to determining, the validity of specific categorical imperatives (general conditional intentions or maxims). The meta-ethical principle that those intentions of the form
We would that anyone did Ai if in Ci
are categorically valid which would be the legislation of an organized community of ideally rational beings qua motivated by the categorically valid intention to maximize the common good, does not absolve us from the necessity to use empirical data in our attempt to determine what ought to be done in particular kinds of circumstance. Any legislator, motivated by the common good, must ask questions of the form: What kind of action in this kind of circumstance would promote the common good? Only an omniscient legislator should not have to hedge his answers with 'probably' and 'for the most part.'

79. Kant is insisting that the principles in terms of which the concept of a categorically valid intention is to be explicated are not empirical principles. They are a priori, and can, in principle, be known by a "mere analysis of the conceptions of morality."16 The fallibility of moral philosophy is not the fallibility of empirical induction.

73. The various so-called formulations of "the categorical imperative" are meta-ethical principles which locate categorical imperatives (in the sense of specific categorically valid maxims) in the total structure of categorically valid intentions to which they belong. Thus the formulation in terms of legislation appropriate to a 'Kingdom of Ends,' though it comes last, actually points to the derivability of categorically valid maxims from the intrinsically categorical intention that "all ends [be] combined in a systematic whole (including both rational beings as ends in themselves and also the special ends which each may propose to himself)."17

74. The formulation "Act on maxims which can at the same time have for their objects themselves as universal laws of nature"18 reflects the logical relation between and and all. The intention -- "from the moral point of view" or, as Kant would say "from the point of view of a rational being as such" -- with respect to anybody that he do Ai in Ci doesn't entail the intention that everybody do Ai in Ci (for we do not intend the impossible). It does, however, entail the wish that everybody did Ai in Ci. Thus, a state of affairs in which everybody conforms to categorically valid maxims is itself a categorical ought-to-be, and the wish that it be the case a categorically valid wish.

75. Finally, the principle that ". . . each [rational being] must treat itself and all others never merely as means, but in every case at the same time as ends in themselves"19 reflects the fact that the intrinsically valid intention which is the prime mover of the domain of categorically valid intentions is the intersubjectively valid intention that each of us rational beings promote our common good. This state of affairs is an end-in-itself in which particular individuals appear symmetrically as agents and patients in an ethical community.


Notes

1 It will be convenient to drop the 'when' and assume, for simplicity's sake, that that action in accordance with such intentions takes place immediately on recognizing the circumstance to be C.

2 This is not to say, however, that

It shall be the case that Tom does A
has no first person parallel which would be subsumable with it under common practical principles. It is merely to emphasize the conceptual primacy of intentions to do even in the case of intentions that someone do.

3 It is, however, a familiar fact that when we become aware of the implications of our beliefs we often change our mind. It is equally true that when we become aware of the implications of our intentions we often, shall I say, change our heart.

4 "Imperatives, Intentions and the Logic of 'Ought'," in Morality and the Language of Conduct, edited by Hector-Neri Castañeda and George Nakhnikian (Detroit, 1963), pp. 173-6; 190-1. Also, "Thought and Action," in Freedom and Determinism, edited by Keith Lehrer, (New York, 1966), pp. 110 ff.

5 This mistake is a consequence of the failure to bear in mind that distinction between what is implied by an intention-content and what is implied by the state of having an intention with that content.

6 This argument clearly presupposes, for its validity, the principle that it is reasonable to do whatever is implied by the content of an intention we happen to have. Explicitly formulated this principle would read

If x has an intention of content I, and I implies (for x) 'I shall do A,' then x ought to do A
and this can be readily seen to be a false categorical imperative.

7 The complexity of the instrumental nomologicals which are relevant to the bringing about in social contexts of any but the most trivial ends must constantly be borne in mind. The simplicity of the schematic letters 'A,' 'C' and 'E,' should not blind us to this fact. We must take into account the effects of our action on the actions of others as well as the effect of the actions of others on the outcomes of our own.

8 Of course, if an intention is merely relatively or hypothetically reasonable, it cannot be categorically reasonable as well.

9 The implications which are formulated in ordinary language as hypothetical imperatives are formulated as 'ought to do's because the consequent intentions are intentions to do:

If one wants . . . , then one ought to do. . . .
On the other hand, the implications which we are now concerned are intentions that something be the case. They appear in the material mode of speech as hypothetical 'ought to be's thus
If one wants X to be the case, Y ought to he the case
e.g.
If one wants a good crop, the soil ought to be moist.

10 It might be thought that since two people can use egocentric referring exprcssions and yet "make the same statement," the same might be true in the case of intentions. But egocentric referring expressions can be counterparts and hence used by different people to makee the same statement (e.g. 'Jones is here by me': 'Jones is there by you'), because they gear in with an intersubjective framework of relative location. If there were intersubjective intentions . . ., but then this is exactly the problem.

11 In the case of wishes as contrasted with intentions, we have the locution

We would that . . .
to contrast with
I would that . . .
Here the 'we' and 'I' express the form of the wish. In the case of intentions, however, it is only intentions to do which exhibit this distinction. The function of the indices is performed in ordinary language by the contrast between "from a personal point of view" and "from ttle point of view of the group" or, of more interest, "from a moral point of view."

12 ". . . this "I ought" (Sollen) is properly an "I would" (Wollen), valid for every rational being, provided only that reason determined his actions without any hindrance" (Immanuel Kant, Critique of Practical Reason and Other Works on the Theory of Ethics, tr. by Thomas K. Abbott. 6th ed. (London: Longmans, Green & Co., 1927), p. 68).

13 Ibid., p. 303.

14 E.g., by Sir David Ross in The Right and the Good (Oxford: the Clarendon Press, 1930), p. 5.

15 Kant, op. cit., p. 125.

16 Ibid., p. 29.

17 Ibid., p. 51.

18 Ibid., p. 56.

19 Ibid., p. 52.