Originally published in Noûs 7 (1973): 179-202.

Actions and Events*

Wilfrid Sellars
UNIVERSITY OF PITTSBURGH

I

1. In his essay, "Sovereign Reason", which gives its name to the book in which it is contained, Ernest Nagel attributes to Brand Blanshard the view that "serious consequences for morals and the life of reason follow from the denial that logical necessity is involved in causal relations" and describes him as arguing that " 'unless necessity does play a part in the movement of inference, no argument will establish anything,' since on the hypothesis that no such necessity exists the distinction between being moved by reasons and being moved by causes is simply an illusion" ([3]: 290).

2. Now, as Nagel points out, a number of quite different considerations are blended together in this way of looking at 'the life of reason.' There is, in the first place, the thesis that causal connection is a species of logical implication or entailment. Nagel to the contrary, I believe that this is a respectable position provided that logical implication or entailment is not identified with the 'subject-matter neutral' implication which, until recently, has been the primary object of attention by professional logicians. Even extreme anti-Humeans have learned from Hume that causal laws are synthetic in the sense that no definitional transformation will exhibit them as instances of subject-matter neutral truths. But it is doubtful that anyone who has held an entailment theory of causality has thought that causal connections were, in the corresponding sense, analytic. And the mere fact that they have referred to causality as a form of logical implication should not be taken to carry this commitment, since the restriction of the adjective 'logical' to analytic implications is an expression of Humean commitment and reflects actual usage only to the extent that it is shaped by teachers of logic.

3. I take it, then, that there is nothing absurd (though much that is controversial) in entailment theories of causation. And, indeed, I have devoted considerable effort over the past twenty years to elaborating and defending such a theory (thus in [6] and in [10]). To engage in this, perhaps quixotic, enterprise is not my purpose on the present occasion, though I have thought it proper to warn the reader that it and related views are closely related to the topics I shall discuss.

4. Next let me note that 'implies' and 'entails' are relation words taking abstract singular terms as their subjects in accordance with the schema:

That-p) implies (or entails) that-q.
This is true whether the implication in question be causal, or logical in that narrower sense to which it has been professionally restricted and to which, for present purposes, I shall also restrict it.

5. Statements to the effect that one proposition logically implies another formulate principles of inference. And it is notoriously important to distinguish the principle of an inference from its premises. Thus, the regimented inference,

p
Therefore, q
has as its premise the proposition that-p, for its conclusion the proposition that-q, and as its principle the higher order proposition:
The proposition that-p implies the proposition that-q.

6. Notice that this schema leaves it open whether or not the principle of inference is a logical truth. Thus, for present purposes, the following schema would be more explicit:

Not both p and not-q
p
Therefore, q.
Here, the premises are the propositions that-(not both p and not-q) and that-p, the conclusion is the proposition that-q, and the principle of inference is the proposition:
that-(not both p and not-q) and that-p imply that-q,
or, if the premises are conjoined into one premise,
that-[(not both p and not-q) and p] implies that-q
.

7. Having offered this finer grained example, however, we can return to the simpler schema, for it suffices for the philosophical points I wish to make.

8. First, a terminological point which is not merely terminological, but a bridge to the point at issue between Blanshard and Nagel. The implication statement

that-p implies that-q
can be rephrased as the necessity statement
that-p necessitates that-q.
In traditional terms, to say that one state of affairs implies another is to say that they are 'necessarily connected,' that the one 'necessitates' the other.

9. Notice that I have dropped the word 'proposition' in favor of 'state of affairs.' My reason for doing so has deep roots in the theory of abstract entities. 'Proposition', in actual usage, is a verbal noun based, I believe, on some such verb as 'propose' or 'propound.' Propositions in this sense are events -- indeed they are actions. They have the form:

(Someone) proposing that-p.
The proposition that-p, in this sense, would not be what ontologists ordinarily have in mind by 'the proposition that-p', but would rather be an event- or action-type which 'involves', in a manner by no means easy to analyze, the proposition that-p in the technical sense of 'proposition' in which this term has come to be used.

10. Just how this technical usage is to be explicated involves reflection on the manifold of distinctions between:

  1. A propounding;
  2. A proposition in the sense of what is propounded;
  3. A proposition in the sense of the physical product of a propounding -- e.g., an ink inscription, a sound, or, more long lived, a trace on a tape recorder;
  4. A proposition in the sense of what is meant by inscriptions and traces;
to which other distinctions can, I am sure, be added.

11. Yet, if I find the term proposition a dangerous one, I am not altogether happy with the term 'state of affairs,' for two reasons: The first is that in ordinary usage states of affairs are temporal, and it would be awkward to rephrase the locution:

the proposition that 2 plus 2 equals 4
by:
the state of affairs that 2 plus 2 equals 4.
The second reason cuts a bit deeper, for the term 'state of affairs' in ordinary usage is connected not with that-clauses but with participial expressions, thus:
Tom's being home for the holidays was a happy state of affairs.
I shall have more to say on this topic later.

12. For something like these reasons many ontologists have been tempted to coin technical terms and to say, for example, that that-clauses refer to 'objectives.' I think that this is a useful term, but having called attention to its availability, and to the considerations which have led to its adoption, I shall continue, for the moment, with the above warnings, to use both 'proposition' and 'state of affairs' in something like the technical senses in which they have been used by ontologists. Needless to say, I do not regard these technical senses as 'clear and distinct.' They are, however, conveniences. We are thrown into philosophy as Heidegger was thrown into the world; the dialogue we continue, we did not begin, and a Whitehead can introduce a whole new vocabulary, and still be understood -- as well as before.

II

13. But enough of this backing and filling. What does this logic chopping have to do with Blanshard's contention that "unless necessity does play a part in the movement of inference" the distinction between being moved by reasons and being moved by causes is simply an illusion?

14. Nagel points out correctly -- though I rephrase what he says in terms of the above distinction -- that from the fact that one state of affairs, that-p, entails another state of affairs, that-q, it does not follow that if the thought that-p occurs to a person, the thought that-q will occur to him. Thus, the necessity which relates to two states of affairs does not carry over into a relation of necessity between the thinking of the one and a subsequent thinking of the other.1 He concludes that even where the relation between two states of affairs is necessary, the relation between the thought of the one and the thought of the other is contingent.

15. Now, this seems obviously to be the case when the thinker in question, as we say, doesn't know that the one state of affairs entails the other. And, clearly, there are many entailments which a given person does not know and many entailments which nobody knows. What, then, if we add the premise that the thinker knows the entailment ? Does this remove the contingency ?

16. Is it a purely contingent fact that Jones, who knows the entailment between the two states of affairs, thinks of the second upon thinking of the first?2 Nagel points out that we can readily think of cases where Jones thinks of the first state of affairs without thinking of the second. Does this suffice to make the relation between the occurrence of the two thoughts,3 when they do both occur, a contingent one ? I shall argue, to the contrary, that it can be a conceptual truth that, other things being equal, the one thought is followed by the other, even though on many occasions the one occurs without the other -- indeed, even if, to push the matter to the edge of paradox, a Jonesean thought that-p is never followed by a Jonesean thought that-q.

17. If I am right, then the way is open to giving something like the 'classical' account of the idea that reasons can be causes. For this account insists that it is because that-p entails that-q that a thought that-p is, ceteris paribus, followed by a thought that-q.

18. It will be instructive to contrast this thesis with Nagel's account of the sense in which reasons can be causes:

A man who first notes a premise A, and then perceives that A logically implies B, is moved by reasons when he accepts B on the evidence of the premise, even if the causal sequence, the thought of A, the perception of the connection between A and B, the assertion of B is a logically contingent one. ([3]: 290; italics mine.)
Three comments are in order:
  1. Nagel implies that to be moved to think B 'by reasons' involves the occurrence of the thought that A logically implies B. Yet, surely, it is only if A's implying B were functioning as a premise that a thought of it would be required; and if it must occur as a premise, then accepting a conclusion for reasons would involve an infinite regress. If it does not function as a premise, the most that is required is that the knowledge (or belief) that A implies B be available.

  2. Nagel speaks of 'accepting B on the evidence of the premise.' But this is just the idea to be explicated. There is more to accepting B "on the evidence of A" than first thinking A and then thinking B. And this more is the heart of the matter. It won't do simply to sandwich in the thought that A logically implies B. For although accepting B on the evidence of A involves the presence, in some sense, to the mind of the fact that A logically implies B, what concerns us is how the entailment functions to make an accepting of B following an accepting of A an accepting of B on the evidence of A. It is the idea that the entailment somehow play a causal role that is the source of the temptation to suppose that a thought of it must occur in the process itself.

  3. By characterizing "the causal sequence, the thought of A, the perception of the connection between A and B, the assertion of B" as "logically contingent," Nagel implies that the concept of a world in which people invariably have thought sequences in accordance with the schema:
    a thought of A,
    a perception of the connection between A and B>
    an assertion of not-B,
    is a coherent one, which is surely not the case.

III

19. An important step in coming to understand how the existence of an entailment between that-p and that-q carries with it the idea that ceteris paribus a person is caused to think that-q by thinking that-p, is to ask what is the difference between thinking that-q upon thinking that-p, where one would characterize this as a mere case of association, and thinking that-q upon thinking that-p, where we are prepared to call this an inference? Note that if, as this question implies, we are prepared to say that inferences of our regimented type are cases of association in which certain additional conditions are satisfied, then we should be happy to say that the premise thought of such an inference causes the conclusion thought. For surely association of ideas is, as critics of Hume have long pointed out,4 a paradigm case of causation, the causing of one mental event by another.

20. I am assuming, it will be noticed, that we are not taking as our paradigm of causation:

One thing (or a group of things) by changing in a certain way causing a different thing (or a group of things) to change in a certain way.
This interventionist model of causation is only one variety of a relation between events by virtue of which the occurrence of the one explains the occurrence of the other. Clearly, we often explain why an object comes to be in one state or to change in a certain way by referring to an antecedent condition of the same object. And even a mechanical explanation need not have the interventionist form, as when, in dealing with a closed mechanical system, one explains its being in a certain state s' at t' with reference to its being in another state, s, at a preceding time, t.

21. Thus we must distinguish between the questions,

What caused Jones to associate B with A?
and
What caused Jones to think B?
The cause of the association may well be a fact about his environment. But given that he has the association, and that no countervailing factors are present, the cause of his thinking B is his thinking A. Furthermore, though his thinking B has a cause, he is not 'caused to think B' in the sense that a thing or person other than Jones himself caused him to think B. One could, if one wished, say that Jones was the cause of his thinking B. But unless this is just another way of saying that Jones thought B, it could only mean that, in the context, Jones' thinking A was the cause of his thinking B.

22. To explain the difference in our regimented cases between inference and mere association, and to understand how, in the case of inference, the causal relationship involves the fact that the premise entails the conclusion -- or even the belief that the premise entails the conclusion -- it will be useful to adopt a Rylean type position which I shall call Verbal Behaviorism.5 According to this reconstruction, 'thinking that-p', where this means 'having the thought occur to one that-p,' has as its primary sense saying 'p', and it has a derivative sense in which it stands for a short term proximate disposition to say 'p'.

23. Notice that I am unabashedly treating that-clauses as quotations in spite of the fact that

Julius Caesar said that the die was cast
is not equivalent to
Julius Caesar said 'the die is cast.'
An adequate account of that-clauses as quotations would require reference to a mode of quotation which spans different languages, as familiar quotation spans the difference between written and spoken forms of the same language. (For an account of such quotations, see my [7]; also [12], Chapter 3.).

24. Verbal Behaviorism, defined as above, is a radical oversimplification. Yet the proper method of rational reconstruction is by successive approximation, and Verbal Behaviorism provides a useful strategy for approaching key philosophical issues in the philosophy of mind; in particular, issues pertaining to perception, inference, and action. Before we can use it, however, we must guard against certain misconceptions which would render it worse than useless. Thus, it is essential to note that just as thinking that-p, in the sense of having the thought occur to one that-p, is not a mental performance, something that one does or could do voluntarily, so, in the Verbal Behaviorist model, saying 'p' is not to be construed as an illocutionary act. It is to be construed, as I have elsewhere put it (for example, in [13]), as candidly thinking-out-loud that-p, and it is not to be confused with 'asserting that-p,' 'telling someone that-p', or any other of the verbal performances so lovingly collected by the late John Austin. Of course, in any ordinary sense, saying 'p' is a performance, because the phrase permits the utterance episode to which it applies to be either a spontaneous thinking-out-loud that-p or a deliberate use of words to achieve a purpose, malicious or benign. I, on the other hand, am using the expression "S says 'p' " in a contrived sense in which these options are closed and the utterance specifically construed as a spontaneous or candid thinking-out-loud.

25. We can imagine a child to learn a rudimentary language in terms of which he can perceive, draw inferences, and act. In doing so, he begins by uttering noises which sound like words and sentences and ends by uttering noises which are words and sentences. We might use quoted words to describe what he is doing at both stages, but in the earlier stage we are classifying his utterances as sounds and only by courtesy and anticipation as words.

26. Only when the child has got the hang of how the sounds function in the language can he be properly characterized as saying 'this is a book,' or 'I shall raise my hand,' or 'it is not raining,' or 'lightning, so shortly thunder,' or 'you spanked me, so you don't love me.'

27. To say what a person says is to give a functional classification of his utterances. The Verbal Behaviorist agrees with Wittgenstein that the meaning of an utterance is its use, but, unlike Wittgenstein, he is careful to distinguish 'use' in the instrumental sense (in which utterances are construed as Austinian performances) from use as function. It is in this latter sense that the meaning, for example, of logical words is their use.

28. Some functional relationships are purely intra-linguistic (syntactical), and are correlated with formation and transformation rules. Others concern language as a response to sensory stimulation by environmental objects -- thus, candidly saying (or having the short term disposition to say) 'lo, this table is red'.6 Still others concern the connection of practical thinking with behavior. All these dimensions of functioning recur at the metalinguistic level in the language in which we respond to verbal behavior, draw inferences about verbal behavior, and engage in practical thinking about verbal behavior -- i.e., practical thinking about thinking.

29. From this point of view, the basic point I wish to make is a simple one. When we characterize a person's utterance by using a quotation, we are implying that the utterance is an instance of certain specific ways of functioning. For example, it would be absurd to say:

Tom said (as contrasted with uttered the noises) 'It is not raining' but has no propensity to avoid saying 'it is raining and it is not raining.'
Thus to characterize a person's utterances by quoting sentences containing logical words is to imply that the corresponding sounds function properly in the verbal behavior in question and to imply that the uniformities characteristic of these ways of functioning are present in his sayings and proximate dispositions to say.

30. But what, even in our simplified model, does all this have to do with the involvement of entailments in the causation of conclusion thoughts by premise thoughts in our regimented cases of inference? The answer is that statements of the form:

that-p implies that-q
are normative statements to the effect that, from a logical point of view, thinkings that-p ought not to be accompanied by thinkings that-not-q and that if a thinking that-p is epistemically sound, then it is properly accompanied by a thinking that-q. These ought-to-be7 statements are essential to the practical thinking which shapes the language learning of the young and keeps entropy away from language once acquired.

31. The functioning which gives the utterances of one who has learned the language their meaning can exist merely at the level of uniformities. When Tom utters or is disposed to utter the sequence: 'p', 'q' where we would not classify his utterances as 'p' and 'q' unless the relevant functioning of these utterances was in his repertoire, we could say that ceteris paribus the former caused the latter and also that the fact that the former is followed by the latter is an instance of a uniformity for which the entailment is responsible -- in the sense that persons who accept the entailment statement have followed the norm it formulates in teaching him the language. In this respect, the linguistic community plays a role importantly analogous to that of the natural environment when it causes association.

32. The child begins at the 'pattern governed'8 level of verbal behavior but subsequently becomes a full-fledged member of the linguistic community and thinks thoughts (theoretical and practical) not only about non-linguistic items but also about linguistic items, i.e., from the Verbal Behaviorist point of view, first level thoughts. At this later stage, he can not only reason in accordance with entailments, he can reason about entailments. And since entailments are principles of criticism, he has now developed from being the object of training and criticism by others to the stage at which he can train and criticise himself and even develop new and more complicated standards in terms of which to guide his own development.9

33. Notice that Verbal Behaviorism permits us to say not only that, ceteris paribus, Jones' thinking that-p caused his thinking that-q (whether this be mere association or inference) but also to penetrate beneath the functional description classification of the utterances (and propensities to utter) by the use of quotation and the devices of indirect discourse, and describe for example, the child as acquiring the propensity to follow one utterance phonemically classified with another utterance phonemically classified. At the verbal level, we can penetrate beneath functional descriptions which, in the case of basic statements and inferences, carry with them causal commitments, to non-functional descriptions of the same utterances which, as non-functional, carry no such conceptual commitment. This is analogous, at a more complicated level, to retreating from:

The burn was caused by fire
to:
The blister was caused by fire.10

34. It is a most significant fact, as I have pointed out elsewhere,11 that the classification of thoughts, construed as classical mental episodes, permits of no such easy retreat to a non-functional level. Roughly, our classification of thoughts, construed as episodes which belong to a framework which explains the kaleidoscopic shifts of sayings and propensities to say, is almost purely functional. We have only the foggiest notion of what kinds of episodes, non-functionally described, perform the relevant functions, though philosophers of a scientific orientation are prepared to characterize them genetically as 'neuro-physiological.' As a result, philosophers unaware of this alternative strategy have the illusion of an ultimacy of the conceptual functioning of thoughts which is responsible for continuing philosophical puzzles about how mental acts are to be fitted into a naturalistic picture of the world.

IV

35. Donald Davidson in his classic paper on "Actions, Reasons, and Causes" ([1]), opens by asking "what is the relation between a reason and an action when the reason explains the action by giving the agent's reason for doing what he did?" He gives the term 'rationalization' a new (and non-pejorative) use by saying that "we may call such explanations rationalizations and say that the reason rationalizes the action." He proceeds to defend "the ancient and common sense position that rationalization is a species of ordinary common sense explanation" ([1]: p. 685). It should be clear by now that I am in general sympathy with this thesis, though I have as yet not even touched on the topic of action.

36. My main uneasiness with Davidson's treatment is that although much of what he has to say is sound, and although he actually puts his finger on the nub of the matter, by failing to discuss the causal character of inference, he leaves a serious gap in his account of how reasons are causes.

37. Davidson puts his thesis in two parts:

C1. R is a primary reason why an agent performed the action A under the description d only if R consists of a pro attitude of the agent towards actions with a certain property and a belief of the agent that A, under the description d, has that property. ([1]: 687).

C2. The primary reason for an action is its cause. ([1]: 693).

38. Now the main source of my discomfort is the expression "pro attitude." Davidson is well aware that this is an omnibus term and lists, in the course of his discussion, some of the specific mental states to which this term applies. Yet he has little to say about this, and the total effect is that when it comes to the mental episodes involved in the causation of action, what he stresses are episodic beliefs about matters of fact.

39. He correctly notes that dispositions and propensities can come and go almost as rapidly as the happenings we take as paradigm cases of causally related events. Thus, he writes "states and dispositions are not events, but the onslaught [and presumably the flight: WS] of a state or disposition is" ([1]: 694). The Verbal Behaviorist would point out that the propensity to say "Damn, I just missed the bus" begins at a certain time in a familiar situation and ceases shortly afterwards, being replaced by somewhat less proximate dispositions to mutter about missing the bus.

40. I pointed out in paragraph 38 that what disturbs me about Davidson's account is that in his examples he tends to stress the onslaught of factual thoughts and leaves the relevant pro attitudes to be relatively long term dispositions which constitute the mental background of the functioning of reasons. Before I elaborate this point, however, let me quote a passage in which Davidson puts his finger on the sort of point I have been trying to make. He is criticizing Melden and, using one of Melden's examples, writes:

A man driving an automobile raises his arm in order to signal. His intention, to signal, explains his action, raising his arm, by redescribing it as signaling. What is the pattern that explains the action? ... is the pattern this: the man is driving, he is approaching a turn; he knows he ought to signal; he knows how to signal, by raising his arm. And now, in this context, he raises his arm. Perhaps, as Melden suggests, he does signal. And the explanation would then be this: If under these conditions a man raises his arm, then he signals. The difficulty is, of course, that this explanation does not touch the question of why he raises his arm. He had a reason to raise his arm, but this has not been shown to be the reason why he did it. If the description 'signaling' explains his action by giving his reason, then the signal must be intentional; but on the account just given, it may not be. ([1]: 692-3).
Shortly afterward, Davidson pinpoints the issue by arguing that:
In order to turn the first 'and' to 'because' in 'He exercised and he wanted to reduce and thought exercise would do it,' we must, as a basic rule, augment condition Cl with:
C2. A primary reason for an action is its cause. ([1]: 693).

41. It is now time to notice that his discussion of the example of signalling quoted above suggests that Melden has in mind an alternative account of the pattern which explains the action, namely that:

[it is] the familiar pattern of an action done for a reason.
He then makes the above point that, in this case,
[the pattern] does, indeed, explain the action, but only because it assumes the relation of reason and action that we want to analyse.
Davidson, as I see it, is here making the same point against Melden that I made against Nagel by calling attention above, paragraph 18, to the fact that the locution "accepting B on the evidence of the premise'' is exactly the one that needs to be explicated in order to cope with Blanshard's problem.

42. Yet Davidson's account, as I see it, is really not much better off, for although he argues, correctly, that causality is the key to a reason's being the reason for an action, he devotes his primary effort not to explicating the mode of causation involved but to defending the idea that the relation could be a causal one against the standard objections offered by philosophers of the Ryle-Melden- Peters stripe. His answers to these objections are sound and to the point. On the other hand, Davidson's constructive account remains undeveloped save insofar as he points out that the onslaught of a disposition is an event and hence that dispositions can serve as causes. Nevertheless, the onslaught he finds to be the key to the signalling example is the driver's noticing or thinking he has noticed his turn coming up. I quote:

To dignify a driver's awareness that his turn has come by calling it an experience, much less a feeling, is no doubt exaggerated, but whether it deserves a name or not, it had better be the reason why he raises his arm. ([1]: 692).
This is immediately followed by a passage in which the role of the pro-attitude, which presumably is a proximate disposition of which the onslaught is equally relevant, is obscured:
The intention with which the driver raises his arm is also not an event, for it is no thing at all, neither event, attitude, disposition, nor object. ([1]: 692).
I am sure that Davidson could have made the same point about the 'belief in which the driver raises his arm. But although both objects of 'belief and objects of 'intention' are queer objects which undoubtedly merit a Quinean regimentation, to treat 'belief and 'intentions' differently in the present context is to muddy the waters.

43. The point I'm making will stand out more clearly if we consider, by contrast, the account which can be given by our Verbal Behaviorist. This account gives, in principle, the cash for the promissory notes12 contained in Nagel's reference to 'accepting B on the evidence of the premise' and Davidson's reference to the essential role of causation in explaining an action in terms of the reason for which it was done.

44. Consider the child who is learning to use sentences which, as we say, formulate intentions; sentences involving an intention-expressing auxiliary verb -- for simplicity, I shall suppose that it is always 'shall.' From the standpoint of non-functional description, it is a matter of learning how to use sentences involving the sound 'shall':

I shall now raise my hand.
Clearly, the child has not learned how to use this sound unless he acquires the propensity to raise his hand, ceteris paribus, upon uttering (or being disposed to utter) the sound 'I shall now raise my hand.' Given that this propensity has been acquired, a necessary condition has been met for redescribing his utterances of the sound '---shall---' as sayings of '---shall---.'

45. But by the same token, the utterance of the sound 'I shall now raise my hand' has become, in the relevant circumstances, "the" cause of his raising his hand. Thus described, the connection is conceptually contingent. When redescribed, however, as a saying of 'I shall now raise my hand,' the connection becomes a conceptual one, for it is a conceptual truth that ceteris paribus a saying or proximate propensity to say 'I shall now raise my hand' is followed by a raising of one's hand.

46. A further condition is that the child acquire the ability to draw practical inferences, that is, inferences which move from practical premises to practical conclusions; thus, and I continue to draw on regimented examples,

I shall do Ay if C.
C.
So, I shall do A;
and, more directly related to the problem of reasons and causes,
I shall bring about X.
Bringing about X implies doing Y in ten minutes.
So, I shall do Y in ten minutes;
where, given that the child has acquired the ability to 'tell time,' to ascribe to him a candid saying or proximate disposition to say:
'I shall do Y in ten minutes'
is to imply that, ceteris paribus, this thought will be succeded by:

I shall do Y in nine minutes.
I shall do Y in eight minutes.
.
.
.
I shall now do Y.

47. To put it bluntly, to say that Jones did Y in order to bring about X, is to say that he went through something like the above practical reasoning. It is in this sense that the 'reason':

to bring about X
is 'the' cause (and it is familiar fact how context-oriented the definite article is) of his doing Y.

48. In short, it is because practical premises can be causes of practical conclusions that reasons can be causes of actions. This is the connection between the dispute about inference between Blanshard and Nagel with which I began and the problem of how reasons can be the causes of actions.

49. Notice that from the standpoint of Verbal Behaviorism, saying (or having the short term proximate propensity to say) 'I shall now raise my hand' is a willing (volition) to raise one's hand. And conceptual connection between willing to raise one's hand and raising it turns out to be a special case of the conceptual connection of thoughts of certain forms with the context in which they occur. Another example, which has already been discussed, is the connection between having a red book present to one's senses and saying (or having the proximate propensity to say) 'Lo, here is a red book.'

50. Obviously, these conceptual connections are not to be confused with the semantical facts that:

'Lo, this is a red book' (said by S at t) is true ≡ there is a red book in a certain place at a certain time,
or that:
'I shall raise my hand' (said by S at t) is realized (i.e., its descriptive content is true) ≡ the person in question raises his hand at the relevant time.
The conceptual connection between the volition and the raising of the hand pertains to causality, yet it is not itself the causal connection; for what is conceptually true is that there is a causal connection between the volition non-functionally described and the raising of the hand; and, again, between the presence of the red book to one's senses and the saying 'Lo, here is a red book' non-functionally described.

51. A fuller account would explore the relations between the various mental states summed up under the omnium gatherum 'pro-attitude' and such immediate practical thoughts as 'I shall now do A' Thus, desires are, roughly, relatively long term dispositions to have thoughts of the form:

(Other things being equal) I shall do A (or bring it about that-p),
where it is presupposed that doing A or contemplating the truth of 'p' is something one would enjoy. (I have explored some of the problems connected with the botanizing of 'pro-attitudes' in [11].) A wish is like a desire, save that the possibility of doing A (or bringing it about that-p) is in doubt. Thinking that one ought to do something or bring about something is like a desire, save that, in a sense which requires careful analysis, the 'point of view' is impersonal. It is along the latter lines that one would defend Prichard's statement that desire and the sense of duty are coordinate species of motive.

V

52. I shall conclude this paper with some remarks on the logic of actions and 'mere' events. As this formulation implies, actions are events. They are events of which the subject, in a sense to be characterized, is a person and are ceteris paribus caused by the person's willing to do them. This implication is built into action words and is reasonably explicit in such locutions as 'bringing it about that...'. Actions have a non-action core which could, in principle, be formulated in ways which abstract from this implication. In addition to this implication of causation by volition, many action words imply a reference to the purpose for which it is typically done. It should be noted that in speaking of the 'non-action core' of an action concept, it is important not to use such phrases as 'one's hand going up.' Obviously, the subject of this expression is not the person but the hand. Thus to formulate the non-action core of the concept of raising one's hand properly, one would have to make it explicit that it is the person that is the logical subject of the event.

53. What is an event? 'Event' is a category expression, and to ask the question is to ask for the place of 'event' in a system of categories. Leaving aside such puzzling occurrences as claps of thunder and flashes of lightning, we notice that events are referred to by singular terms which have, to use a technical locution, 'propositional form' and, more specifically, both a logical subject (or subjects) and a corresponding verbal expression; thus:

Caesar's crossing the Rubicon.13

54. Event expressions, being propositional singular terms, remind us of that-clauses, which are also propositional singular terms, and it is tempting to assimilate the two. And, indeed, we say:

That Caesar crossed the Rubicon is true,
as well as:
Caesar's crossing the Rubicon took place.
Are these not intimately related? Each is a statement in which the subject is a singular term and the verb is in the third person singular, thus: 'is [true],' and 'took place'. Obviously, not all that-clauses correspond to event expressions (thus, 'that 2 plus 2 equals 4'). But can we not regard event expressions as a proper subset of that-clauses? I think the answer is yes; but in defending it, some explanation must be given of the fact that that-clauses contain tensed verbs, whereas event expressions contain participial expressions. Thus, in sentences to the effect that a certain event took place, typically the event expression itself contains a present participal but is followed by such full-fledged tensed verbs as 'is taking place/ 'took place' and 'will take place,' as in the above example.

55. If we are to regard event expressions as that-clauses, then we must construe event expressions as they occur in the primary contexts we have been considering as that-clauses in which the main verb has the peculiar form illustrated by:

That Caesar crosses the Rubicon in 56 B.C.
Notice that we are prepared to say, somewhat wistfully,
That Nixon ends the war in 1970 will be true.
(One is struck by the appropriateness of the phrase 'narrative present.') On the other hand, we are less likely to say:
That Caesar crosses the Rubicon in 56 B.C. was true,
preferring instead,
That Caesar crossed the Rubicon in 56 B.C. is true,
and, if we were forced to use the predicate 'was true/ we would say:
That Caesar crossed the Rubicon in 56 B.C. was true.

56. Temporal that-clauses and event expressions play different roles, but, I suggest, they can fruitfully be assimilated, provided these differences are taken into account. My suggestion, then, is that we reconstruct:

Caesar's crossing the Rubicon in 56 B.C. took place
to read:
That Caesar crosses the Rubicon in 56 B.C. was true.
This account carries with it the idea that (and I continue to leave aside interlinguistic considerations) both formulations be construed as referring to past use-occasions of the English sentence:
Caesar crosses the Rubicon
and as characterizing these past use-occasions as true.
Nixon's ending the war in 1970 will take place
would, correspondingly, characterize future (1970) use-occasions of the sentence:
Nixon ends the war in 1970
as true. Hopefully by 1971 we will be in a position to say:
Nixon's ending the war in 1970 took place,
in which case we will be referring to then past use-occasions of the sentence:
Nixon ends the war in 1970,
and be saying of them that they were true.

57. In brief, this reconstruction gives us a unified grasp of apparently unrelated modes of discourse by assimilating event expressions to that-clauses, and the predicates 'is taking place,' 'will take place,' and 'took place' to the straightforwardly semantical predicates 'is true,' 'will be true,' and 'was true.' I realize, of course, that those who regard using a tensed copula with the predicate 'true' as a sin against the Holy Ghost will not be persuaded.

58. I might put my thesis by saying that events are a species of proposition. I am not unhappy about this way of putting it, provided we bear in mind the ambiguities of the term 'proposition' to which I alluded above, for, as we saw, propositions as propoundings or proposings are themselves a species of events.

59. The most frequent objection I encounter when defending the view that events are a species of proposition is that propositions are 'abstract' entities, whereas "what could be more concrete than an event!"14 The proper answer is that what ontologists call 'abstract' entities are more properly called 'conceptual' entities. For although historically such things as triangularity and justice, which are in a straightforward sense abstract, have been taken to be paradigm cases of conceptual entities, not all conceptual entities conform to those paradigms. This stands out most clearly if, as I believe we must, we take individual concepts seriously. It would be paradoxical in the extreme to say that in the context:

Socrates is a substance,
where 'Socrates' stands for an individual concept (see [15]), what it stands for is an 'abstract entity.' Or, to use a somewhat less controversial example of a Fregean type, in the context:
Jones believes that Socrates was captious,
the word 'Socrates' stands for an individual concept -- that is, in a suitable sense, a conceptual entity. But, again, it would be paradoxical to say that, even in this context, 'Socrates' stands for an abstract entity.

VI

60. The generic form of events, sentences, and, hence, of action sentences is:

S's V-ing { took place
is taking place
will take place

I have proposed that this generic form be reconstructed as:

That S Vs { was true
is true
will be true

As for the finer grained form of event sentences, the key, as I see it, has been given us by Professor Romane Clark, whose unpublished paper, "In Any Event: Davidson's Analysis of the Logical Form of Action Sentences," caused so many of the ideas with which I was groping to fall into place, that I rarely have read a paper with greater excitement.15

61. Clark correctly calls attention to the fact that to understand the logical form of statements involving event expressions, we must understand the logical form of those more basic statements which, without the use of event expressions, tell us that something has taken place. Thus, in my terms, to understand the form of:

S's V-ing took place,
we must understand the structure of:
S V-ed.
In terms of a by now classic example, to understand the structure of:
(The event of) Jones' buttering the toast in the bathroom at midnight for Smith, while the band was playing in honor of Georges Pompidou, took place,
we must understand the structure of the corresponding first level statement:
Jones buttered the toast in the bathroom at midnight for Smith while the band was playing in honor of Georges Pompidou.

62. I threw in the 'while'-clause, because I have long been groping for an account of the basic use of such words as 'while,' 'after,' 'before,' which does not treat them as relation words. For to treat them as such forces us into an event ontology. Thus, if in their basic use they are relation words, they require singular terms of which to be predicated, and these singular terms would be event expressions, construed as basic, for surely 'while,' 'during.' etc., must occur in ground floor descriptions of the world. (See, for example, [9], esp. pp. 550-2, 567-77.).

63. Leaving aside the 'while'-clause, why is not the remainder of the sentence to be construed as asserting a five term relation to hold between Jones, the toast, the bathroom, the time, and Smith? Clark, to my mind, uses the correct strategy -- although his formal apparatus may need some reshaping -- by construing 'in the bathroom,' 'at midnight,' and 'for Smith' as adverbial modifiers of the verb 'to butter.' Adverbial modifiers of this kind (though this does not seem to be true of all adverbs) make, in effect, big verbs out of little ones, just as logical connectives make big sentences out of little ones. Thus, the second sentence above, in spite of its complicated surface structure, would still have the form:

V-ed (Jones, the bread).
The 'bit' verb construed by the use of these modifiers remains a dyadic one, in spite of the involvement of the bathroom, the time, and Smith. Looking at it more closely,
Jones buttered the toast for Smith at midnight in the bathroom
can be reconstructed as:
[In the bathroom [at midnight [for Smith]]] buttered (Jones, the toast),
which, given carefully formulated rules (implicit in ordinary usage), entails:
Jones buttered the toast,
Jones buttered the toast for Smith,
Jones buttered the toast at midnight,
Jones buttered the toast in the bathroom,
not to mention various combinations of the same ingredients, all of which, properly regimented, begin,
Jones buttered the toast... .

64. Clark's analysis, in my opinion, also provides the key to the problem of the identity of events and hence of actions. This problem, which any theory of events must face, is at the center of the stage -- not only in action theory, but, for example, in recent discussions of the mind-body problem. But that is, again, a story for another occasion.


References

[1] Donald Davidson, "Actions, Reasons, and Causes", Journal of Philosophy 60 (1963): 685-700.

[2] Romane Clark, "Concerning the Logic of Predicate Modifiers", Nous, 4 (1970): 311-35.

[3] Ernest Nagel, Sovereign Reason and Other Studies in the Philosophy of Science (Glencoe, Illinois: Free Press, 1954).

[4] Wilfrid Sellars, "Some Reflections on Language Games", Philosophy of Science 21 (1954): 204-28.

[5]---------, "Empiricism and the Philosophy of Mind", in Minnesota Studies in the Philosophy of Science, Vol. I, edited by H. Feigl and M. Scriven (Minneapolis: University of Minnesota Press, 1956): 239-52.

[6]---------, "Counterfactuals, Dispositions, and the Causal Modalities", in Minnesota Studies in the Philosophy of Science, Vol. II, edited by H. Feigl, M. Scriven, and G. Maxwell (Minneapolis: University of Minnesota Press, 1957): 225-308.

[7]---------, "Abstract Entities", Review of Metaphysics 16(1963): 627-71.

[8]---------, Science, Perception, and Reality (London: Routledge and Kegan Paul, 1963).

[9]---------, "Time and the World Order", in Minnesota Studies in the Philosophy of Science, Vol. Ill (Minneapolis: University of Minnesota Press, 1963): 527-616.

[10]---------, "Induction as Vindication", Philosophy of Science 31(1964): 197-231.

[11]---------, "Thought and Action", in Freedom and Determinism, edited by K. Lehrer (New York: Random House, 1966).

[12]---------, Science and Metaphysics (London: Routledge and Kegan Paul, 1963).

[13]---------, "Language as Thought and Communication", Philosophy and Phenomenological Research 35(1969): 506-27.

[14]---------, "Metaphysics and the Concept of a Person", in The Logical Way of Doing Things, edited by Karel Lambert (New Haven: Yale University Press, 1969): 219-52.

[15]---------, "Towards a Theory of the Categories", in Experience and Theory, edited by L. Foster and J. W. Swanson (Amherst: University of Massachusetts Press, 1970): 55-78.


Notes

* This paper was originally presented at a symposium on the topic at the University of North Carolina, November, 1969.

1 I am quite aware that I am using thoroughly regimented examples of inference. A botanizing of the thought patterns which can appropriately be characterized as inferential would require the space of another essay. It should not be forgotten that the dialectics of the controversy has developed in connection with such regimented examples and, in my opinion, rightly so. As far as I can see (although this might not have been the case), no essential changes in the overall strategy of the argument would be involved in taking less tidy cases into account.

2 Notice that his knowledge that the one entails the other should not be construed as the employment of an additional premise -- for, as has often been pointed out, most brilliantly by Lewis Carroll, the principle of this inference would not be the original entailment but a more complicated one to the effect that, roughly,

that-(that-p is the case) and that-(that-p entails that-q) entail that-q.

3 The regimented inferences with which we are dealing constitute only one of the many patterns of thought which are governed, in the sense to be clarified, by entailments.

4 Though the use they have made of this point can be criticized.

5 For an elaboration of the strategy of using Verbal Behaviorism as a model to clarify issues in the philosophy of mind, see my [13].

6 I use the contrived operator 'lo' to indicate that the context is one in which the sentence which follows it is functioning as a perceptual response to an environmental object, and to indicate that this context is preserved in the retention of the sentence in the 'memory system' of the perceiver. In this respect, 'lo' functions in perceptual contexts much as 'shall' will be used to function in contexts pertaining to practical reasoning.

7 For an exploration of the distinction between 'ought to be' and 'ought to do' -- where the 'do' ranges over actions proper -- see [13].

8 This term was introduced in an earlier attempt to understand how rules are causally involved in the generation of meaningful verbal behavior, [4], reprinted as the concluding chapter of [8].

9 For a theory of induction built around this interpretation of entailment see [10]; see also [4].

10 Compare [1].

11 Most recently in [14]. For an early formulation, see [5], especially sections XII-XIV.

12 The reader will, of course, be able to distinguish between such cash 'in principle' and a promissory note.

13 Perhaps lightning flashes and claps of thunder can be included in the general category of singular terms having propositional form by taking into account those strange sentences which involve the pseudo-subject 'it,' which have intrigued many philosophers; thus:

It thundered,
to which would correspond the singular term 'it's thundering'. In the case of lightning, the candidates would be the sentence 'There was lightning' and the singular term 'there being lightning.'

14 Needless to say, in this context, events must be carefully distinguished from event types.

15 This paper has since been published as Section I of [2]. -- Editor's note.