Originally published in Freedom and Determinism, edited by Keith Lehrer (New York: Random House, 1966): pp. 105-39
My aim in this essay is to explore the interconnection of a number of concepts which are of particular interest to the philosophical psychology of action. The interconnection of these concepts concerns their respective roles in the concept of practical reasoning, and it is with some reflections on the latter that this essay will begin and end. By “practical reasoning” I understand reasoning which, if carried to its conclusion, ends in a practical assertion in one’s heart and, possibly, on one’s tongue. By a practical assertion, I mean in the first instance an assertion which is of the form
such and such shall[1] be the case.
I use the very uninformative placeholder “such and such”—which amounts to a variable for states of affairs {106}—because part of my task is to determine what sort of expressions can appropriately replace it, i.e. what sort of states of affairs are appropriately referred to by a practical assertion.
The above stipulation involves many promissory notes, not the least of which concerns the fact that the phrase “practical reasoning” is normally construed to cover not only such reasonings as that expressed by
I shall drive this nail,
Therefore, I shall get my hammer,
but also reasonings which culminate, not in practical assertions of the narrow kind described above, but rather in assertions to the effect that such and such an action ought to be done, or that such and such an object or state of affairs is or would be good. And, indeed, I would agree that such reasonings are cases of practical reasoning. It will therefore come as no surprise when I propose the thesis that, in their primary use, “ought” and “good” are special cases of “shall.” If I am right, the above definition of practical reasoning is not as narrow as it looks, and an exploration of the structure of reasonings which culminate in explicit “shalls” is capable of throwing light on reasoning which pertains to obligations and values. But, for the moment, this more general application must remain a promissory note.
It is clear that the above line of thought is akin to the imperativist interpretation of practical reasoning developed by, among others, R. M. Hare and Hector Castañeda. It differs from them primarily in stressing that practical reasoning, like theoretical reasoning (in the classical sense in which it is contrasted with practical reasoning), need not be addressed to an audience. It can, and usually does, “go on in one’s head.” An imperative is essentially something that is addressed to somebody, telling him or her to do something. Thinking is, I believe, fruitfully conceived as “inner speech,” though it would be a mistake to construe all “inner speech” as “talking to {107} or addressing-oneself.” I may address myself in inner speech—as when I think “Watch out, Sellars, you nearly split an infinitive.” Thus, while we can address imperatives to ourselves in inner speech—tell oneself to do something—the thought expressed by
I shall drive this nail
should not be construed as though it would be more perspicuously expressed by
Myself, drive this nail!
Yet, for all this, there is much in common between the fundamental position I am taking and that of the imperativists. They conceive of a practical statement as either an imperative or a statement which entails an imperative. I conceive of a practical statement as either a shall-statement or a statement which entails a shall-statement.
Shall-statements express intentions. To explain exactly what this means, let me introduce some notions from the Philosophy of Mind. First of all, let us distinguish between a thought and its overt expression in language. By “a thought” I mean a mental episode, something that occurs in a mind at a certain moment.[2] Thoughts, as I am using the term, are examples of “mental acts.” But the term “act” is misleading, as we shall see, for it is easily confused with action. Thoughts are characterized by reference to the utterances in which they would eventuate if the thinker were in a thinking—out—loud frame of mind. If, now, we draw on the familiar distinction between an episode and a disposition or propensity, we can characterize a belief as a disposition to have thoughts of a certain kind. Thus, to believe that LBJ is President is to be disposed to have (in appropriate contexts) thoughts of the “LBJ is President” kind.
{108} Similarly we must distinguish between three meanings of “intention”
or, though not actually thinking this, he may be disposed to have such thoughts. Just as a person can be said to believe something throughout a certain period of time, even though he is not constantly thinking the relevant thought (indeed, it may not occur to him), so a person may be said to intend something throughout a period of time, even though the appropriate shall-thought does not occur to him. The classical doctrine that actions are typically initiated by volitions has recently come in for some hard knocks; I have defended this thesis, and shown, to my own satisfaction, at least, that the major types of objection to it are based either on category mistakes or on too narrow a conception of empirical meaningfulness. According to the account I there offered, we acquire what rational psychology calls the “faculty of volition” in the course of learning the language of intention or “shall-talk.”
There is an important similarity between learning to make the language-entry transition[3] of responding to presented red objects by saying “This is red,” and learning the language-departure transition, which joins the saying of
{109} I shall now do A
and, consequently, the thought
I shall now do A
with a doing of A. Until a child has acquired has acquired this connection, he has not learned the meaning of “I shall now do A,” and until he has learned this, he cannot learn the full meaning of “shall,” for all other uses of “shall” owe their connection with action to their connection with this use of “shall.”
According to the proposed model, a volition is a thought of the form
I shall now bring about X (or avert Z) by Y-ing,
where Y-ing is something, as we say, within the agent’s voluntary control, and where, in the limiting case of minimal actions done for their own sake, “bring about X (or avert Z) by Y-ing” collapses into “do Y.” The above formula is designed to avoid the mistake of supposing that the content willed is limited to actions that are under the agent’s voluntary control. But though not limited to them, volitions must include them. Somewhat more artificially, we can represent a volition as having the form
I shall now do A in order to bring about (or avert) B,
noting that its contents may even be more complicated, thus
I shall now do A in order to bring about (or avert) B in spite of bringing about or preventing C.
What differentiates a volition from an act of intending, is (a) its now character; and (b) the fact that the central place in what is intended is something (putatively) doable here and now.
A simple case of the relation of intending to volition can be illustrated by considering Jones, who has formed the intention of raising his hand in ten minutes. Suppose {110} that no alternative course of action recommends itself to him. Then we may picture the situation as follows:
I shall raise my hand in ten minutes
⋮
I shall raise my hand in nine minutes
⋮
I shall raise my hand now
(which culminates in action, if Jones happens not to be paralyzed).
Intentions, like beliefs pertaining to particular matters of fact, involve an apperceptive framework of orientation in a spatio-temporal world. That one’s “place” in time is constantly and systematically changing is, of course, an essential feature of this framework; one that is reflected in—indeed constituted by—a systematic change in the content of thought with respect to tense, temporal connectives and the like.
But the fact that certain shall-thoughts are such that, in the absence of paralysis, they culminate in action, is only one, if an essential, feature of the framework of shall-thoughts or intentions. Two other major features remain to be noted. They concern, roughly, the existence of underivative intentions, and the mode of derivation of one intention from another. The latter topic is cousin to the familiar topic of imperative inference, which concerns the conditions under which an imperative sentence may be validly derived from a given set of sentences. In our framework, it is the problem of the circumstances under which a shall-conclusion can be validly arrived at from a given set of premises.
The first principle is easy. It parallels Poincaré’s principle that no imperative conclusion can be drawn from premises which do not include an imperative. Once this has been agreed upon, however, it has proved easy to get bogged down in controversy concerning what the other principles should be. Fortunately for our purposes, it turns out that only one principle is needed. It goes
{111} If P implies Q, then Shall [P] implies Shall [Q].
The one word that causes trouble, of course, is the little word “implies.” I have in mind the ordinary sense in which implication is the relation which authorizes inference. I shall assume that we know our way around in this ordinary sense of “implies,” and that it involves no paradoxes. I shall further suppose that there are at least two species of implication in the ordinary sense: (a) logical, and (b) physical or natural; finally, that there is such a thing as derivative implication, where something implies something else, relatively to a given assumption.
Notice that according to the above principle a shall-implication has the form
Shall [P] implies Shall [Q];
and it is true if and only if P implies Q. If so, then, all steps in practical reasoning would have to be, as reconstructed in this framework, derivations of one Shall from another. There would be no mixed derivations, such as:
Shall [if it is raining, I will come in]
It is raining
Therefore, Shall [I will come in].
Fortunately, it can be shown that this pattern of argument can be reduced to our paradigmatic form. For, relatively to the assumption that it is raining,
I will come in, if it rains implies I will come in,
just as relatively to the assumption that Socrates is a man,
All men are mortal implies Socrates is mortal.
Thus, relatively to the proposition that it is raining,
Shall [I will come in if it rains] implies Shall [I will come in].
In short we admit as valid those practical reasonings the premises of which combine an intention with factual information, provided that, relatively to the factual information {112}, the content of the premise-intention implies the content of the conclusion-intention.
Notice that no meaning has been given to “not shall [P]”, as contrasted with “Shall [not-P]”. Shall-sentences, unlike ought sentences, do not have contradictories. It is particularly important, in this connection, to distinguish first-person present-tense statements of the form “I intend to do A,” or “I intend that S shall be P,” from simple expressions of intention. Of these two statement forms, the latter (and, less perspicuously, the former) stands to the simple expression of intention “S shall be P”, (“I shall do A”), much as “I believe that S is P” stands to “S is P.” Unlike shall-sentences, sentences of the form “X intends that S shall be P” have two negations:
X intends that S shall not be P
and
X does not intend that S shall be P.
These considerations throw additional light on the principle which relates “P implies Q” to “Shall [P] implies Shall [Q],” for they commit us to the position that the latter implication does not have a contapositive. If it is countered that an implication which has no contapositive is not an implication, the answer is that this simply reflects the extent to which the theory of implication has been dominated by the study of reasoning concerned with mathematical relationships and empirical matters of fact.
The analogy drawn above between
X intends that S shall be P
and
X believes that S is P
calls to mind the problem of the “logic” (which might better be called the “axiomatics”) of belief, and raises {113} the corresponding problem of the logic of intention—where this means not the logic of shall-sentences, but rather, the logic of, intention-sentences. The latter is a richer topic, the study of which is scarcely under way. Two remarks will be helpful in connection with latter stages of the argument: (1) We must distinguish between
I.P implies Q ⊃ Shall [P] implies Shall [Q]
and
II.P implies Q ⊃ (X intends that Shall [P]) implies (X intends that Shall [Q]).
The latter principle is, as stated, false. Only if radical restrictions are placed on the way in which P implies Q, as by limiting it to the case where seeing that P implies Q is a necessary condition for understanding these propositions, can it be true. (2) If such a restriction is introduced, II becomes a special case of principle
III.X thinks that P implies Q ⊃ (X intends that Shall [P]) implies (X intends that Shall [Q]);
or, equivalently,
IIIa.X thinks that P implies Q ⊃ (X does not intend that Shall [Q] implies (X does not intend that Shall [P]).
Before I turn to the central part of this essay, which concerns the relations between intention, desire, enjoyment and action, a further group of distinctions is necessary. The first concept to be defined is that of preference. The concept, as I introduce it, is a thin one; for it makes no reference to enjoyment or satisfaction, but is defined solely in terms of intendings or shall-thoughts. A preference is a disposition pertaining to such thoughts. Roughly,
{114} X prefers A to B
implies that where A and B are thought to be relevant, but incompatible (not jointly realizable), X on asking himself,
Shall [A · ~B]? Shall [B · ~A]?
will intend
Shall [A · ~B].
It should, of course, be noted that
X prefers A to B
is compatible with
X prefers B · C to A · C,
which tells us that, in relevant circumstances, X will intend
Shall [B · C · ~A]
rather than
Shall [A · C · ~B].
A more complete expression of this intention, given that the answer to the previous questions is as above, would be
Shall [B · C in spite of ~A].
Here is the place to supplement our previous remarks on the logic of shall-sentences by noting that there is no logical move from
Shall [A], Shall [B]
to
Shall [A · B],
let alone
{115} Shall [A] · Shall [B].
No sense has been given to the latter, nor, if an attempt is made to appeal to the peculiar sense (“conjunction introduction”) in which
P, Q
‘implies’
P · Q
has any sense been given to “Shall [P, Q]”.[4]
We can introduce the concept of indifference to a kind of state of affairs with respect to another kind as follows:
X is indifferent to G with respect to K
has the sense of
X prefers neither G · K to ~G · K nor ~G · K to G · K.
Finally, we can say that
X is favorably disposed to L
has the sense of
X prefers L to ~L,
which tells us that X will intend
Shall [L · ~L], i.e. Shall [L]
rather than
Shall [~L · ~L], i.e. Shall [~L].
It is important to note that not all choices are expressions of preference, a fact which requires a refinement of {116} the above definition. The point is an important one in moral psychology, although to elaborate it would take us beyond the scope of the present paper, because it brings out into the open a widespread confusion between character and nature. Thus, if a person queries
Shall it be the case that X · ~Y? Shall it be the case that Y · ~X?
and answers
It shall be the case that X · ~Y,
it is presumably true of S at that time that, if he were to query thus, he would answer thus—from which, together with the above definition, it would follow that all choice is an expression of preference.
But preferences and traits of character in general are, not just true hypotheticals about intendings. They are hypotheticals defined in the vocabulary of practice or conduct, which are such that the primary mode of establishing them is induction within this conceptual framework.[5] Thus, even if invisible cerebroscopes and a knowledge of psychophysical connections enable invisible Martian interlopers to formulate true hypotheticals about which of two alternatives a person would choose, if he were to deliberate, they would not have discovered, ipso facto, a preference or character trait. All choice may, in a suitable sense, be the expression of the nature of a language-using animal. But it is certain that not all choice is an expression of character. Choices form character as well as express them.
Just how the above definition of preference should be amended to meet this point is hard to say. Perhaps something along the following lines would do the job:
Given that intentions can be derived from other intentions along the lines explored at the end of part I, what is the status of underived intentions? In other words, what is the status of intentions that have not been reasoned from other intentions? Roughly, what is it to desire some thing for its own sake? I say roughly, because the word “desire” has in ordinary usage a more specific meaning than the technical usage I have in mind. Yet a provisional explication of this technical usage will help us build the framework in terms of which the more general concept can be understood. This strategy is supported by the fact that, since the time of Plato, philosophers have tended to use the term “desire” in something like the technical sense that I have in mind.
It might be thought that the interesting question about desires is “how do desires get us to act?”—which they obviously do. Do desires cause volitions? Actually this question is the least interesting of the questions we can ask about desire; for, on the account which shall give, the connection between desire and volition is a logical one. This follows from the fact that desires are relatively long-term dispositional intentions. Thus their relation to action is a matter of the fact that, relatively to acquired information intentions imply and can be seen to imply other intentions, and indeed, on those occasions in which practical reasoning reaches its proper conclusion, {118} intentions pertaining to action, and hence, where the time of action is hic et nunc, volitions.
But to say that the desire that-p is a disposition to have thoughts of the form
Shall [that-p]
or, perhaps, that to desire a state of affairs is to be favorably disposed toward it, in the sense defined above, leaves unexplained the fact that, in addition to having something to do with action, desires also have something to do with pleasure or satisfaction. I believe that the best model for understanding this relationship is to construe desires as mental needs on the model of bodily needs. The key theme here is the fact that the “filling”—to use Plato’s metaphor—of a need brings with it (ceteris paribus) pleasure.[6] Thus, if “X desires that-p” could be explicated as “X needs to believe that-p,” the connection between desires and “pleasure” would be a special case of the connection between needs and “pleasure”; and is it not plausible to say that “X desires to be loved” implies “X needs to believe that he is loved” and “X would be pleased if he believed himself to be loved?”
But even if we ultimately conclude that desires are of the nature of needs to believe, the perspective of the preceding paragraphs precludes any quick move in this direction, for we have construed them as dispositional intentions. It may, however, be the case that a closer examination of this construction will narrow the gap between “disposition to intend” and “need to believe.” Desires clearly resemble intentions in the sense that they are capable of being realized. Realization in this sense is a concept akin to truth. Intentions—and volitions in particular—are realized if and only if the state of affairs intended {119} comes to exist. Similarly, a desire is realized if and only if the state of affairs desired comes to exist. And just as a judgment can be true without being known to be true, so, in this technical sense, an intention or a desire can be realized without being known to be realized.
But for a desire to be realized in this sense is obviously not the same thing as for it to be satisfied. Desires seem to have a logical connection with enjoyment or satisfaction, which is not accounted for by anything we have said so far about intentions and their realization.
We seem to be confronted by a dilemma. If we define a desire as a dispositional intention, we acknowledge a logical tie between desire and action, but seem to leave no place for a logical tie between desire and enjoyment or satisfaction. If, on the other hand, we define desire in such a way as to establish a logical tie between desire and satisfaction (along lines hinted at above), we seem to lave no place for a logical tie between desire and action. Yet prima facie desire is logically tied to both satisfaction and action.
In order to clarify the problem, we must explore the concept of the satisfaction of a desire. As has already been indicated, this requires a discussion of the concept of enjoyment, i.e. of the context
X enjoys Y-ing.
The first point to he made is a familiar one. To enjoy Y-ing is not to do two things—i.e. to Y and “feel pleasure.” This phenomenological point was appreciated by Aristotle, and misused by Berkeley. The misuse is instructive. Berkeley argued that a loud noise is a pain and drew the conclusion that, since pain is a mental state, loud noises must be mental states, and hence (by considerations of continuity) all noises must be mental states. The argument is, of course, a bad one. The major premise should have been not “a loud noise is a pain” but “hearing a loud noise is a pain, i.e. painful.” And from this the desired conclusion does not follow. Yet Berkeley had, like {120} Aristotle, seen that the painful character of, the hearing of a loud noise is not a matter of doing two things, hearing and feeling; but rather doing one thing in a certain manner, i.e. hearing painfully.[7]
Taking the contexts
X enjoys Y-ing
X disenjoys Y-ing
as phenomenologically ultimate, and limiting ourselves, for simplicity’s sake, to the former, we turn next to the distinction between conceptual and non-conceptual enjoyments. Obviously thoughts can be enjoyed as well as muscle movements. Yet to draw the distinction is to recognize that, since distinctively human activities involve conceptual as well as non-conceptual components, the extent of purely non-conceptual enjoyments is less than might initially be thought. Indeed, in so far as Y-ing is not a minimal action, but extends beyond a bodily or mental motion, and is a bringing about of something that is not in our immediate voluntary control, to enjoy Y-ing is to enjoy the thought that one is Y-ing.
Now the satisfaction of a desire obviously has something to do with the belief on the part of the person who is doing the desiring that the desired state of affairs obtains. This suggests that
X desires that S be P
has a strong conceptual tie with
X is disposed to enjoy the thought of S being P.
{121} It will be remembered that in Principia Ethica, G. E. Moore finds an intimate connection between the idea of a certain state of affairs being a pleasant idea and the desire that the state of affairs obtain. Moore interprets the connection, however, as a causal one, a link in a longer chain which can be represented (with the necessity qualifications) by the schema
(i)X finds the idea of a certain state of affairs pleasant(ii)X desires the state of affairs in question(iii)X (thinking that doing A would bring about this state of affairs) wills to do A
This picture presents the relations between desire and enjoyment, on the one hand, and desire and action, on the other, as causal. Our problem is to see whether we can so reconstruct the situation that both relations are logical or conceptual.
If we were to define desire in a broad sense, (which includes, for example, wishes) as a disposition to enjoy the idea of a certain state of affairs obtaining (and, perhaps, to disenjoy the idea of its not obtaining), then we would have a strategy for explaining the conceptual connection between desire (in that narrow sense in which it is contrasted with wishing) and enjoyment. Here the enjoyed idea of a state of affairs obtaining would be an enjoyed belief that the state of affairs obtains. To speak of a desire in this sense as satisfied may well imply that it is also realized[8] and hence that the belief enjoyed is true. But it would at least imply that the relevant belief was enjoyed—if only for a moment.[9]
{122} But such a definition destroys, at least in appearance, the logical connection we feel to exist between desire and action, a feeling which found expression in our previous definition. We would seem to be forced to construe the connection between desire and volition as, in a broad sense, a causal one. The other horn of the dilemma, it will be remembered, was that to define the desire that p be the case as some form of disposition to intend that p be the case would, even though it recognized the logical tie between desire and action, fail to account for the prima facie connection between desire and satisfaction.
Is there any way in which both connections can be accounted for, so as to give them at least an element of conceptual necessity? It could, of course, be done in a trifling way, by defining desire in terms of a simple conjunction, a disposition to enjoy thinking that p is the case and a disposition to intend that p be the case. But it is clear; that what we are looking for is a logical connection between these two dispositions, by virtue of which they belong together and the conjunction is not arbitrary. Is there some mediating connection that we have overlooked?
What about a possible connection between
X is disposed to enjoy Y-ing
and
{123} X is disposed to Y?
That some such connection seems to exist is clear. Is it logical or causal? Is it a logical or a causal fact that, other things being equal, people do what they enjoy doing? Of course, given a certain description of what a person is doing, it may be true that he does not enjoy doing that. Thus a person may eat shredded wheat for breakfast and not enjoy what he is doing qua eating shredded wheat.[10] But might it not be the case that there is another description of what the person is doing, which preserves the desired connection?[11]
Now I am strongly inclined to think that there is a conceptual connection between doing (not doing) and enjoying (disenjoying). I do not have anything particularly helpful to say on this subject. It does, however, seem to me that, if there is such a connection, then the strategy to be employed in connection with the problem I have posed becomes clear.
Thus, if we assume that the above principle is true, and apply it to the case where the doings in question are thinkings, we get the derivative principle
X is disposed to enjoy thinking ‘it shall be the case that-p’ implies that S is disposed (ceteris paribus) to think ‘it shall be the case that-p.’
If, therefore, we can find a connection between
{124} X is disposed (ceteris paribus) to enjoy thinking ‘it is the case that p’
and
X is disposed (ceteris paribus) to enjoy thinking ‘it shall be the case that p,’
we would have succeeded in finding a mediating link between
X is disposed (ceteris paribus) to enjoy thinking ‘it is the case that-p’
and
X is disposed (ceteris paribus) to think ‘it shall be the case that-p,’
and, therefore, between desire as the sort of think that can be satisfied, and desire as the sort of thing which, ceteris paribus, finds its expression in action.[12]
But what of the additional hypothesis, introduced in {125} the previous paragraph, to the effect that there is a logical connection between
X is disposed (ceteris paribus) to enjoy thinking “it is the case that-p”
and
X is disposed to enjoy thinking “it shall be the case that-p”?
Notice that there is a subtle tense difference between the thinkings involved in the two dispositions. But if we suppose, for the sake of the argument, that tense as such is transparent to enjoyment, i.e. that, ceteris paribus, if, at t, X enjoys thinking “it will be the case that-p,” then at the appropriate time (t + Δt) X enjoys thinking “it is now the case that-p,” and vice versa, then we can focus our attention on the connection between
X is disposed (ceteris paribus) to enjoy thinking ‘it will be the case that-p’
and
X is disposed (ceteris paribus) to enjoy thinking ‘it shall be the case that-p.’
But if this is the connection on which the argument hinges, then we are in the neighborhood of an old friend, the problem of the relation between intention in the ordinary sense and belief; the problem, in other words, of the relation between intending to do something and believing that one will do it.[13]
Now I think it would generally be agreed that intending to do A is not the same thing as believing that one {126} will do A. If they were the same, then our question would already have been decided in the affirmative, and the argument could stop right here. On the other hand, granted that they are not identical, there is nevertheless a very close relationship between them. For intending to do A involves knowing that, unless one is paralyzed and unless unfavorable circumstances prevent one’s minimal action from developing into a doing of A, one will do A—that is, of course, unless one changes one’s mind, before the time of action occurs. Thus, while intending to do A does not consist in believing that one will do A, it essentially includes it. How is this to be understood?
Notice that coming to intend to do A neither requires a prior belief that one will do A, not is, it incompatible with the existence of such a belief. One may believe on inductive grounds[14] that one will, in all probability do A and then ask oneself, “After all; shall I do A?” and deliberate whether or not to do it.[15]
As I see it, the key to an understanding of the involvement of belief in intention is a rejection of Hare’s neustic-phrastic analysis of the linguistic expression of intentions and beliefs. According to this type of analysis, it will be remembered, the expression of an intention (which Hare mistakenly models on imperative performances) has the form
S shortly being P, please.
Thus reconstructed, it has the phrastic “S being P” in common with the corresponding expression of belief, which joins to it a different neustic and is reconstructed as
S shortly being P, yes.
Now it is clear that to take this line is to remove the commitment to its being the case that S is, shortly P from both intendings-out-loud and the covert mental episodes for which they are the model. As I see it, on the other hand, thinkings-out-loud that S is P are not to be analyzed as a matter of “S being P” plus an assertion sign or affirmative nod. Rather the participial phase “S being P” is to be understood as a sterilized or “bracketed” form of “S is P,” for use in larger contexts where the commitment to its being the case that S is P to be suspended.
And, as I see it, the context provided by that in the verbal expression of intention which makes it an expression of intention does not suspend this commitment. If we take the word “shall” to be the specific sign of intention, then I suggest that if any reconstructing is to be done, we reconstruct
S shall be P,
I shall soon do A
as
Shall [S will be P],
Shall [I will soon do A]
—i.e. as involving not sterilized phrases such as “S going to be P” or “I doing A shortly,” but the corresponding full-blooded indicatives. On this interpretation, an intending is more than but includes a thinking that something will be the case, just as being under the visual impression that S is P includes but is more than a thinking that S is P.
But surely, it might be said, one could agree that
S thinks “I shall shortly do A”
entails
S thinks “I will shortly do A”
(not, of course vice versa), or even, more generally, that
S thinks “it shall be the case that-p”
entails
S thinks “it will be the case that-p”
without granting that
S enjoys thinking “it will be the case that-p”
entails
S enjoys thinking “it shall be the case that-p”;
or, to focus the issue decisively, and presupposing what we called the transparency of tenses to enjoyment, without granting that
S is disposed to enjoy thinking “it is the case that-p”
entails
S is disposed to enjoy thinking “it shall be the case that-p.”
The counter I must offer, if my thesis is to stand up, is that “shall” is also transparent to enjoyment. Is there any reason to suppose this? Certainly, if one makes a confusion between an expression of intention and the corresponding {129} autobiographical statement which ascribes to oneself that intention, thus between
I shall shortly do A
and
I intend to do A shortly,[16]
then the thesis will seem to be false. Roughly I may approve of doing A, but when the question is raised; I may well not approve of the fact that I so approve. Again, I may enjoy the thought of doing A, but may well not enjoy the thought that I enjoy this thought. This is a special case of the general point that where φ is a predicate applying to states of affairs Σ, it does not follow from the fact that someone enjoys thinking that Σ obtains, that he enjoys thinking that φ(Σ) obtains. But to all this the proper rejoinder is that “shall” is not a predicate. It is a concept word only in that broad sense in which words which play essential roles in reasonings can be said to express concepts.The logical “or” is also not a predicate, yet it also expresses a concept. On the other hand, “or” and “shall” differ in this key respect: that, whereas the former pertains to the content of thought and is essential to the logical space of predicates, “shall,” in spite of its logical role, can be said to be a manner rather than a content of thought. This is the element of truth in the notion of “shall” as a neustic, and explains why, in spite of its logical role, it is most misleading to say that “shall” expresses a concept.
The central role of “shall” in the rational reconstruction of practical reasoning, and in the logic of ends, means, policies and principles (topics which have been adumbrated in the first section of this essay and which will be further developed in the next section) is the guiding {130} thread of the entire argument.[17] For the moment, however, I must limit myself to the bare assertion that when we enjoy thinking something, the enjoyment is a function of the conceptual content of the thought in the narrower sense of content into which “shall” does not enter. If, in this sense, “shall” is transparent to enjoyment, and if there is the postulated connection between enjoying thoughts and thinking them, then the thesis that desire, defined as a dispositional intention, has a logical connection with both satisfaction and action would be established.
Let me emphasize once again that not all current uses of the term “desire” can be explicated along these lines. As far as I can see, the term is sometimes used in such a way that desires constitute a subspecies of the sort of thing I have been talking about, so that it would remain to spell out the differentia. Sometimes, on the other hand, the term is used to refer to what I would call “hungers” or “appetites”—i.e. dispositions not to intend, but to enjoy or disenjoy Y-ing. But the botanizing of mental states and dispositions is a task of great complexity which, at a sophisticated or critical level, remains largely to be done. Many attempts along these lines have failed because they were based on mistaken theories about the fundamental categories of mentalistic discourse, thus “philosophical behaviorism.”.
It will be useful to return to the topic of practical reasoning with an artificially simple case. Let us suppose a situation in which a certain end E can be realized by bringing about state of affairs M1 which in its turn can be brought about by performing minimal action A1. Let us also suppose that there is only one available alternative course of action, initiated by minimal action A2, which would prevent the occurrence of E, by bringing about M2. Finally let us assume that, to the person doing the deliberating, everything but E is indifferent.
From an analytical point of view, two stages can be distinguished in the deliberation. The first lays out the basic implications involved. Indeed we have, according to our assumptions, two pairs of equivalences:
E if and only if M1
M1 if and only if A1~E if and only if M2
M2 if and only if A2
Hovering in the background, we have the intention which finds its expression in
Shall [E].
The second stage of deliberation might be called the summative stage. In it the various features of two alternative courses of actions are brought together logically to be evaluated as wholes. Thus the above equivalences generate the two summative implications:
A1 ≡ E · M1 · A1
A2 ≡ ~E · M2 · A2
Given the assumption made above that everything mentioned but E is indifferent, it follows that our deliberator prefers Σ1 to Σ2 (where Σ1 represents E · M1 · A1 and Σ2 represents ~E M2 · A2)—i.e. that he is disposed to think
Shall [Σ1 · ~Σ2]
rather than
{132} Shall [Σ2 · ~Σ1].
It should also be noted that relative to the equivalences specified in the first stage,
~Σ2, i.e. ~(~E · M2 · A2)
is necessarily equivalent to
(~~E) · ~M2 · ~A2.
Thus the next stage of the deliberation can be represented as the categorical intention
Shall [E · M1 · A1 · (~~E) · ~M2 · ~A2].
It is particularly important to note that it is the complex intention which has just been analyzed, rather than the simpler intention
Shall [E],
which serves as the major premise in the practical syllogism in which the deliberation culminates.
The form of this syllogism is a simple one, which rests upon the fact that
‘Shall [X · Y]’ implies ‘Shall [X]’.
Thus the “syllogism” is simply
Shall [E · M1 · A1 · (~~E) · ~M2 · ~A2]
Therefore, Shall [A1].
Although E can properly be said to be the end of the action in which this deliberation culminates, ‘Shall [E]’ is not the major premise of the relevant practical syllogism. The importance of this point stands out clearly in more complicated cases of deliberation. As our second example, let us consider a case which differs from the above in that M1 and M2 are not indifferent, but positively valued.
The first stage of the deliberation remains as before, as does the summative stage. It is when preference turns {133} one of the envisaged incompatible courses of action into an intention that the difference appears.
Let us suppose that although E, M1 and M2 are all positively valued by the deliberator, he prefers
~E · M2 · A2
to
E · M1 · A1
This means that, in the absence of any restructuring of the situation, the deliberation will culminate in a practical syllogism, of which the major premise is
Shall [~E · M2 · A2 · ~E · ~M1 · ~A2],
and which has as its conclusion
Shall [A2].
If, to tie up our present discussion with our early remarks on volition, we suppose that this conclusion has, from the standpoint of a finer-grained analysis, the form
Shall [I will do A2 in ten minutes],
then, ceteris paribus, this intention, which can be characterized as a decision, will, by virtue of meshing in with the agent's sense of temporal location, develop into the volition
Shall [I will now do A2],
which volition is the initial stage of doing A2.
We have often been warned against thinking that the end (in the sense of end-in-view) of an action is that which comes last in its envisaged scenario. In avoiding this trap, one might make the opposite mistake of supposing that the end of the action is the state of affairs specified by what we have found to be the major premise of the practical reasoning which culminates in the action. This, however, would not correspond to the way in which we actually use the term “end.” The latter seems rather {134} to stand for the combination of states of affairs ingredient in the scenario which carries the burden of the preference.
Thus the end (or purpose) of doing A2 in the second example is to bring about M2. If we add to the latter phrase the qualification “in spite of preventing E and M1” we have what is often called the intention of the action, but “intention” in this sense must be distinguished from a broader (and technical) use, in which it refers as well to states of affairs to which the agent is indifferent, but which he believes to be involved in the relevant implications. In the latter sense, the intention of the agent can he described as that of bringing about ~E · M2 by doing A2, rather than bringing about E · M1 by doing A1.
The above analysis permits a brief aside on the logical relation of
X desires that-p
to action. It suggests that, whatever a more complete analysis would yield,
X desires that-p
logically implies
if X believes that he can bring about, in a way to which he is indifferent, M, to which he is indifferent, and which is the immediate necessary and sufficient condition of its being the case that-p, and X prefers no available alternative course of action, then X intends the realization of M.
A third example adds relatively little that is new, but it is perhaps worth considering because it prepares the way for an exploration of more realistic cases. Suppose this time two “ends,” E1 and E2, such that
E ≡ M
M1 ≡ M2
M2 ≡ A1E2 ≡ M3
M3 ≡ M4
M4 ≡ A2
where A1 and A2 are incompatible. {135} The deliberation is sparked by the questions ‘Shall [I will bring about E1]?’ ‘Shall [I will bring about E2]?’ These questions are themselves generated by the thoughts ‘Shall [E1]?’ and ‘Shall [E2]?’, together with a recognition of the impossibility of realizing both of the states of affairs in question.
This time the summative stage is represented by the questions
Shall [E1 · M1 · M2 · A1 · ~E2 · ~M3 · ~M4 · ~A2]?
Shall [E2 · M3 · M4 · A2 · ~E1 · ~M1 · ~M2 · ~A1]?
That is, schematically,
Shall [Σ1 · ~Σ2]? Shall [Σ2 · ~Σ1]?
Suppose that the preference is Σ1; then
Shall [Σ1 · ~Σ2]
becomes available to serve as a premise from which follows the conclusion
Shall [A1].
Since in this context, the person deliberating thinks that E2 implies A2 and, indeed, that ~A2 implies ~E2, and since the major premise of the practical syllogism implies not only
Shall [A1]
but
Shall [~A2],
the reasoning
Shall [~A2]
Therefore, Shall [~E2]
is available. It is the immediate availability of this reasoning which supports principle IIIa (p. 113 above). Needless to say, when the period of deliberation is over and the implications of E2 are out of mind, ‘Shall [E2]’ may well tend to reassert itself. We may then well desire in abstracto {136} that which we have earlier decided to forego in favor of an incompatible alternative.
The next type of case I shall consider but only by way of indicating broad lines of strategy, concerns action on policy. The fundamental point I wish to make is that it is essential not to assimilate action on policy to using means to achieve an end; Thus, one might think that, in action on policy, the fact that
On all C-occasions I will do A
logically implies
On this C-occasion I will do A
makes my doing A on this C—occasion a logically necessary means to the end of my doing A in every C-occasion. There is, indeed, such a thing as intending to bring about series of like events. Thus, I might want to win every one of a series of games; to bring about this end, winning each game is a logically necessary means. But this is not an example of action on a policy.
As a first approximation, we can say that a person acts the policy of doing A when he is in C, if, with respect to every situation which promises or threatens to be of kind C, he is disposed to think
Shall [if I am in C, I will do A].
Notice that the universality comes into the description of the agent’s propensity rather than into the content of the particular intendings which it is his propensity to have.
Notice, again, that the intendings which manifest a policy are underivative intendings. That is to say, the propensity in question is not that of coming to think
Shall [if I am in C, I will now do A]
as a conclusion from another intention.
The implementation of policies involves practical reasoning, but the intention which expresses the policy is {136} not itself the conclusion of practical reasoning. It is essential not to confuse the fact that
Shall [I will espouse policy P],
which expresses the intention to acquire a certain propensity, can well be the conclusion of a practical inference, with the mistaken idea that the intention which defines the policy must therefore be derivative. A means-end argument can justify
Shall [I will espouse policy P],
without requiring that the policy intention
Shall [If I am in C, I will do A]
be derivable from another intention in accordance with an implication of the form
(Shall [X]) implies (Shall [If I am in C, I will do A]).
Thus the difference between an intending which constitutes a whim and one which expresses a policy is not that the former might begin with
Shall [If I am in C, I will do A],
as an underived intending, whereas the latter would begin with
Shall [All of my C-occasions will be occasions on which I do A]
and move thence to the conclusion
Shall [If I am in C, I will do A];
the difference is rather (at least in first approximation) that, although the intending is the same in each case, the whim intention does not manifest a general propensity to have underived intentions of this kind.
A simple case of a conflict between achieving an end and carrying out a policy can thus be represented as follows:
Shall EShall [If I am in C, I will do A2]Relative to being in C:Relative to being in C:E ⊃ M, M ⊃ A1(If I am in C, I will do A2) ≡ (I will do A2)I am in CI am in CE ≡ E · M · A1(If I am in C, I will do A2) ≡ (I will do A2)Σ1 = E · M · A1Σ2 = (If C, then A2) · A2
The summing-up stage can be represented as:
Shall [E · M · A1 · ~(if C, then A2) · ~A2]?
Shall [(if C, then A2) · A2 · ~E · ~M · ~A1]?
If the person in question does A1, we would regard this is a good reason for saying that, when the chips were down and he was in C, he preferred bringing about E by bringing about M by doing A1 to doing that Which conforms to this policy in that kind of circumstance.
We have arrived at the point where, to say anything more of interest, one would have to say a great deal. I have in mind the general topic of action on principle. While the latter grows naturally out of the topic of action on policy, it requires an exploration of the relation of the forms of thought in terms of which individuals believe and intend, to the fact that they are members of a community. This is a large and difficult area which far transcends the limits of this essay. I have barely scratched the surface on another occasion,[18] and the following remarks are intended only to indicate the general character of the relevant distinctions as they relate to the problem at hand. The theme is an Hegelian one in modern dress, a transposition into another key of his concept of Objective Spirit. It concerns the relation of “we” to “I” and postulates a form of thought which is intending as one of us or, as I put it in the essay referred to above,
Shallwe [X],
{139} as contrasted with
ShallI [X]
The point is not that groups can, in a legitimate sense, be said to intend. It is the more radical one that individuals can intend sub specie communitatis. Thus, and individual can have an intention of the form
Shallwe [X].
And the members of a group can have common intentions not only by virtue of the fact that each thinks sub specie individualitatis
ShallI [——],
but by virtue of the fact that each thinks sub specie communitatis
Shallwe [——].
“I” and “we” come into the form as well as the content of intending. We need an expression that is related to “We intend that —” as “shall” is recognized to be related to “I intend that —.”
I shall not attempt to carry this through on the present occasion. The essential point is that after “we”-intentions have been botanized in terms of such notions as policy, means-end, etc., and the concept of practical reasoning has been expanded; to include them, it turns out that the concept of preference must also be expanded to include not only the dimension in which policies conflict with ends, but that in which a person’s answer to the questions
ShallI [Σi · ~Σj]?
ShallI [Σj · ~Σi]?
breaks with or coincides with
Shallwe [Σi · ~Σj].
[1] Let me remind the reader that in this and other essays I use “shall” in such a way that it always (as it does not in ordinary English) expresses an intention—in a suitably broad sense to be characterized below—of the speaker.
[2] The word thought is also used in the sense of what is thought. In the latter sense a thought is, for example, a proposition which can be thought by many minds or by the same mind at different times.
[3] Wilfrid Sellars, Science, Perception and Reality (London, 1963), pp. 328ff.
[4] It might be claimed that I am inconsistent in denying sense to ‘~Shall [P],’ and ‘Shall [P] · Shall [Q]’ while admitting ‘Shall [P] implies Shall [Q]’. But the latter, involving the relation word ‘implies’ which is not a truth-functional connective, is in the meta—language, and mentions rather than uses the shall-sentences.
[5] This does not mean that a preference for X over Y must be inferred only from past choices of X as against Y. That a person prefers X to Y may reasonably be inferred from his physiological make-up or childhood training.
[6] A bodily need, such as that for calcium, requires, of course, no awareness that anything is lacking. Mental needs, on the other hand, clearly involve awareness in one way or another—just how, remains to be seen. Yet they have this in common with bodily needs, that we can have them without knowing that we do.
[7] This phenomenological fruit of conceptual analysis, which emphasizes the adverbial character of painfulness and enjoyment is, needless to say, compatible with a non-phenomenological, theoretical, account of these verb-adverb situations, which interprets them as psychophysical fields, the describing of which involves the use of neither verbs nor adverbs. It is the feeling that some such account must ultimately be given which has often led to the conviction that enjoyment and unenjoyment are to be analyzed as composites of non-hedonic and hedonic states.
[8] Otherwise one might say that the person merely thought that it was satisfied.
[9] Whether the person gets any other enjoyment if the desire is realized in fact, as well as in belief is, in many cases, a matter of the causal impact on him of the desired state of affairs. It is, of course, conceivable, in these cases, at least, that even if a person’s desires were always realized, the world might be so arranged and his ignorance of causal relationships so profound that the only satisfaction he ever got was an ephemeral conceptual enjoyment of believing that the states of affairs desired were realized. Where on the other hand the object of desire is to engage in an immediately enjoyable activity, the connection between a desire’s being realized and the existence of satisfactions other than that of believing the desired state of affairs to exist is considerably closer. It might be added that, at the very moment of coming to believe that the state of affairs in question obtains, one may see that it involves features to which one has a strong aversion, so that this belief is not even momentarily enjoyed. But we are speaking of propensities which are manifested only ceteris paribus.
[10] For the sake of completeness we must make at least a passing reference to disenjoyment. The corresponding question would concern the character of the connection between
X is disposed to disenjoy Y-ing
and
X is disposed not to Y.
[11] Notice that it is not being suggested that everything a person does falls under an action-description which specifies something he is (ceteris paribus) disposed to do. We have already seen that not every action is an expression of character.
[12] Notice that, if the point made on p. 120 is correct, then in addition to the above mediating connection between enjoying thinking that p is the case and intending that p be the case, there would be a direct connection between “S enjoys Y-ing” and “S enjoys thinking ‘I am Y-ing,” and hence, by virtue of the connection between “S enjoys Y-ing” and “S is disposed to Y,” and between “S is disposed to Y” and “S is disposed to intend ‘I shall Y,’” there would be the desired connection between “S is disposed to intend to Y” and “S is disposed to enjoy thinking ‘I am Y-ing.’” If we replace the verb “to Y” with “to bring it about that p is this case,” the connection becomes that between
X is disposed to think “I shall bring it about that p is the case”
and
X is disposed to enjoy thinking “I am bringing it about that p is the case.”
This, however, would at best be a special use of the connection for which we are looking, for one can clearly desire to bring it about that p is the case while being indifferent to the idea of p’s being them case as such.
[13] The problem is often formulated in terms of the relation between intending to do A and “predicting” that one will do A. But it is clear that predicting is a performance which can occur in the absence of belief, or, to put it in terms of out earlier discussion, to publicly predict that p will be the case is not as such a case of thinking-out-loud that p will be the case. It is therefore, a mistaken model for thinking covertly that p will be the case.
[14] On the other hand, coming to intend to do A is logically incompatible with prior existence of such a belief, if that belief is described as grounded in the decision to do A.
[15] No ‘metaphysics’ of scientific determinism can contradict the conceptual truth in the framework of persons and what I call the practical modalities—that one could always have done something other than what one did, and, hence, that one can always do something other than what it is highly probable that one will do.
The idea that scientific determinism entails fatalism rests, as I see it, on the confusion between
It could not have been the case, relative to the antecedent state of the universe, that S did A,
which involves the modality of scientific explanation, with
S could not have done (was unable to do) A, although he was neither paralyzed nor were the circumstances unfavorable,
which involves the practical modality “can do” (or “is able to do”).
[16] That this form of words can, and often does, serve the purpose of expressing the intention is granted.
[17] In the above argument as elsewhere in this essay, I have attempted to apply to the practical aspects of mind the neoclassical framework developed in “Empiricism and the Philosophy of Mind” in Herbert Feigl and Michael Scriven, eds., Minnesota Studies in the Philosophy of Science, Volume I (Minneapolis: University of Minnesota Press, 1956). Initial steps were taken in this direction in “Imperatives, Intentions and the Logic of ‘Ought’” in Methods Vol. III (1956), 227-268; reprinted with substantial revisions in Hector—Neri Castañeda and George Nakhnikian, eds., Morality and the Language of Conduct (Detroit: Wayne State University Press, 1963).
[18] “Imperatives, Intentions, and the Logic of ‘Ought’,” in Hector Castaneda and George Nakhnikian, eds., The Language of Conduct (Detroit: Wayne State University Press, 1963).