Elizabeth Anscombe and the Classical Will: An Examination of Section 29 of Intention 

by  Steve Bayne


[ADDENDUM APRIL 19, 2009: This, brief, paper was written in 2005. Since that time, I have considered serious revisions and extended discussion. Within a couple of months, I will have completed a full length manuscript on Anscombe's Intention. This larger work covers in great detail her work in philosophical psychology. It is titled Understanding Intention: Elizabeth Anscombe's Philosophical Psychology. In this book I undertake to examine the historical underpinnings and significance of her work. Among the authors discussed in connection with Anscombe's views are Ryle, Davidson, James, Bradley, Moore, Reichenbach, Chisholm, Kenny, Melden and numerous others. This book should be released within the year 2009.]
In section 29 of _Intention_ Elizabeth Anscombe addresses a dilemma. It is arrived at in the course of examining a question she had raised earlier: How do we know our intentional actions? The question itself is in need of clarification, and while she appears almost deliberate in not providing a clear answer I will maintain that, if the question is going to be resolved at all, we must recognize a distinction between volitional actions and acts of will. Anscombe's examples of intentional action include actions performed on impulse (arm raising) as well as actions far more deliberate, such as her example of someone operating a pump and thereby poisoning a cistern from which intended victims are expected to take their water (sections 23-26). Besides resolving the dilemma knowledge of our intentional action presents, another advantage to drawing the distinction I hope to make is that Anscombe's disparate examples of intentional actions are easily accomodated within a single theory. I will consider in some detail how this distinction is to be made and why, but first we need to seize hold as firmly as possible the dilemma of section 29.

The dilemma Anscombe uncovers is not a dilemma for her theory, if for no other reason than that she does not really offer a theory. Rather, it is a dilemma for those who hold certain plausible assumptions about how we know our actions. I may know my intentional action of opening a window, e.g., by observing my hands as they move and seeing the window opening. At the same time I nonobservationally know the action as my intention to make the window open. The result of what I do with my hands may be thought of as part of the intentional act itself. Inasmuch as it is willed, if I am to know it as part of my intentional action I must know it as willed. But this part of my action, says Anscombe, can be willed only in one way, such as when I stare at something and will it to move. The problem with this is that in fact it simply does not occur in connection with opening the window.

If we attempt to avoid the difficulty by identifying what we "do" with what "happens," then we are faced with the dilemma that we know the action both by observation and by some sort of introspective acquaintance. But two knowledges implies there are two things, and so identifying the two is an absurdity; that is, if I identify the two things then I am placed in the very circumstance which, as we have just seen, compels us to distinguish them in the first place.

Crucial to Anscombe's alleged dilemma is the issue of what willing consists in, if anything. Unlike Anscombe who says very little on this matter, it was an issue that defined what I shall call the "classical theory of the will." While there is very little that all classical theorists agreed upon, there is one thing that they did have in common, and here I mean to include James, Bradley, Stout, Alexander, and Shand - among others. What they all had in common was the centrality of the "idea" in understanding the will.

Perhaps the most the most complete statement of the role of the "idea" in classical theory was a series of articles by F. H. Bradley in MIND (1902, 1903, 1904) in which the author attempts to bring to completion views he had developed earlier in reaction to James. Central to Bradley's theory was the idea that "A volition is the self realization of an idea with which the self is identified." (MIND n.s. 1902. in CE. p. 477) One respect in which Bradley's position can be viewed as a deepening of the issues raised by James (Principles of Psychology vol. II. Dover. pp. 486-5920) is the place occupied in the theory by the "self." I will have occasion to raise this as an issue a bit later, but there were other theorists, now largely forgotten, who contributed to the discussion, and whose ideas must either be recalled or rediscovered if a viable theory of the will is to be achieved. G. F Stout was one such theorist, as we shall soon discover.

Anscombe takes to task, as did a number of her contemporaries, those who would include ideas as constitutive in some sense of willful actions. The position Anscombe has taken up to this point can best be thought of as one where there are in a sense only primitive or "basic" actions involving the limbs. These are the only real actions (Davidson follows in this). Here there is no hint of a need to introduce "ideas," such as those James had relied on nearly three quarters of a century prior in accounting for basic bodily actions. For Anscombe, all but these primitive actions are accretions of context, supervening upon such primitive actions. Let's take a closer look at whether an act of will can be analyzed or in some sense regarded as consisting in two components: will and result. It is out of consideration of these two aspects of action that Anscombe discovers the dilemma which is at the center of controversy in section 29 of Intention.

When we throw a ball it is easy enough to regard the action as including the movement of our arm, and the bringing together of our fingers tightly around the ball, and finally, letting go after moving the arm forcefully. We are not so so much inclined to include the movement of the ball through the air as part of the action as we are to regard it as being included in the result of that action. But how much are we to include in the action, that is, the "real" action? When I write, the movement of the pencil is difficult to include with certainty in either the action or its result. Stout's view was that it belongs with the action itself.

Similarly in writing, so long as the pen does not sputter or otherwise become unmanageable, not only are the motions of the hand, but also those of the pen included in what I regard as my own action in distinction from its results. I say *I* write, not that I make my pen write. (Studies in Philosophy and Psychology. Macmillan. 1930. p. 85)

Once we accept this distinction between the act and its result, we can go on to ask how we know each. Both what we have been calling the "primitive" or "basic" act, as well as the result, are intended; that is, to know our intentional action we must know both the action and its result as intended. But the problem Anscombe raises is this: I cannot know the result of the action as something willed by way of observation. Anscombe takes the position that the only sense that can be attached to willing something to happen is where I stare at it and will it to move; but this doesn't happen with respect to the result of moving the pencil as I write. This sort of willing, as Anscombe carefully notes, may be involved in attempts at telekinesis, say, but not in the willing that is involved in intending to write a letter. However, if we attempt to avoid the problem this presents, viz., of how we know the result as an intended part of our action, we encounter considerable difficulty.

Again, one way of escaping the problem would be to include the result as part of the action itself. But if we do so it would then appear that there are two ways of knowing the action. First, by observation and, secondly, in the special way she had earlier spoken of by which we know the position of our limbs without looking. Again, however, we are faced with the difficulty that two knowledges imply two things known, and so identifying the two is an absurdity. This creates a dilemma, or so Anscombe maintains.

Imagine that I engage in the intentional action of moving a matchox by reaching out and giving it a gentle nudge. Falling into the dilemma, Anscombe alleges, follows from what she calls a "mad" account. What the advocates of such an account do is first detach one part of the willed action, in our case the nudging, from its result (the motion of the matchbox) and then say that the first part I know non-observationally while the second part I come to know by observation, although both are "willed in the intention." But this assumes that the will distributes over the two parts of the action and that both parts can be described as "willed" in the same sense. But Anscombe as we have seen objects. I may "will" a matchbox to move, but I do not will to make the nudging motion with my finger, say, in the same sense, at least, of 'will'. Is there something we have missed? I suspect there is, and what that something is, if I am right, is the distinction between volition and will - a distinction few of the classical theorists recognized, something Anscombe shares with them.


In resolving Anscombe's dilemma I have already suggested that a distinction will have to be maintained between 'volition' and the 'will'. Proceeding to the details of how this is to be achieved requires that we begin with at least an intuitive understanding of what qualifies as a volition (or being volitional). This will be accomplished to some degree by way of illustration. My aim is to offer a solution which combines a distinction Anscombe makes together with certain elements introduced by the classical theorists (James, Bradley, Stout, Shand, Alexander, etc.) Even so, explaining 'volition' by illustration is most effective when these considerations are not present or at best merely implicit.

Suppose I try to move my hand and my foot moves; I then try to move my foot and my hand moves; so to get my foot to move I try to move my hand but, then, my other hand moves, and so on. Compare this to the ordinary case where when I try to move my hand my hand moves, and so on. Volition, as I shall use the term, is what these two cases have in common; in both cases there is "voluntary" movement. The cases differ, however, in the matter of will. In the first case the will is never satisfied; in the second it is. If the metaphor helps, consider the possibility that volition may have remained the same with satisfaction of the will varying across worlds. Not all philosophers would agree. Philosophers, such as Locke, who hold that volition is just the actualization of a disposition or "power" would most likely take exception to this. My knowledge of the common component is knowledge of one sort; my knowledge of the difference between them is knowledge of another sort. We don't come to know the common component by observation, but we can come to know by observation that the will is satisfied by looking to see that the intended action is carried out.

This distinction between volition and will, at least in principle, enables us to avoid complementary errors committed by two classical theorists (James and Bradley), on the one hand, and Anscombe (and most likely Davidson) on the other. The mistake James and Bradley made was to make the will depend on the "idea" in all cases. In so doing, they were compelled to conclude that even primitive arm raisings, even those which Anscombe would describe as being done for no reason at all, involve acting on an idea. This seems to require more of the "idea" than it can deliver. Correspondingly, Anscombe goes to the opposite extreme, suggesting that none of what are properly regarded as my actions depend on acting on an idea. This is not to say that there is no sense of "idea" Anscombe finds useful or meaningful, only that ideas as she understood them were not entailed by intentional actions. It was this effort to purge ideas from actions, however, that led her to the opposite extreme, for now the only real actions are basic or primitive actions - the rest being merely 'contextual', something the classical theorists would have thought of as a poor substitute for ideas. What is being suggested here is that we can avoid the dilemma Anscombe identifies at the onset by recognizing a distinction between volition and the will and by so doing steer a desirable middle course between the extremes of Anscombe and the classical theorists and their complementary errors.

Acts of will are here taken to be those actions which depend on ideas, but only after taking into account two important considerations. The first of which are Anscombe's arguments against the classical theorists. The second being that distinction she comes very close to making explicit between an intentional action and an action done "with" an intention (I enclose 'with' in double quotes to indicate a special sense of 'intention', one we have already discussed in some detail elsewhere). We shall take it that actions depend on an idea when they depend on actions performed "with" an intention, such as an arm raising might. No longer need there be the suggestion that arm raisings, themselves, depend on an idea. Nor could we deny that some actions depend on an idea; these and these alone would be actions performed "with" an intention. Anscombe's aversion to "inner movements" obscures the relation between her view that we arrive at some "actions" by layering contexts over primitive actions and her view that what we do we may do "with" a certain intention. This difficulty is largely overcome by acknowledging a class of actions which themselves require actions performed "with" an intention. Before the only bono fide actions were "basic" or "primitive" actions, actions which do not require context in order to be actions. Indeed, it is consciouness of being performed "with" an intention that led some to say, perhaps not fully understanding why they said it, that "willing is attention to the end" (Samuel Alexander on James, 1911). An action, such as arm waiving for no reason, is "volitional" where there is only attention to the act of willing which brings it about in contrast to opening the window - where what I pay attention to is the windows opening, which is hardly attention to what brings it about.

A volitional act is one intiated by a person. A willful act is a volitional act performed "with" an intention. I will have occasion to mention persons only once more, and this in connection with "negative actions." It's role, however, is central to the classical theorists but has been largely neglected since Wittgenstein. This mistake is something which must be overcome if we are ever to achieve a theory of action in any way complete.


Unless we are dealing with extraordinary circumstances, I would never say, "Look, the matchbox is moving as my hand touches it; I must be moving the matchbox!" But neither can I know the matchbox moved without observation. If moving the matchbox is what is *done*, can't I know what I am doing in two different ways; first, in that seemingly special way perhaps like that where I know how my arm is positioned without looking (but with a "text"), and second, in the way I commonly know what is done by simply observing what has happened? The objection Anscombe makes, that an affirmative answer involves the account she deems "mad," does not presuppose one way or the other whether what might properly be called the volitional aspect of the action be identified with some other component feature than what is called its "result." Opinion on this matter is far short of unanimous. Let's take a look at a couple of the possibilities available to classical theorists. While Locke held that (Essay. vol. 1. Dover p. 313) volition is just the exercise of the will, others have held that volition is a state preceding the act. (e.g., Alexander Shand "Types of Will" MIND. 1897. p. 290) Stout, however, took a quite different position, one which came under the scrutiny of some of the finest action theorists of his time, saying

A volition is a desire qualified and defined by the judgement that, so far as in us lies, we shall bring about the attainment of the desired end. "Voluntary Action" [1896] in Philosophical Studies in Psychology 1930. p. 53.

Bradley, however, in a detailed criticism argued that the expression "so far as in us lies" can be understood in either the sense of "so far as in us lies physically" or "so far as in us lies psychologically," and that even in the latter case there is more than one possible reading. More radically, Bradley rejects the need for either judgment or desire as essential to the will (compare here the Ducasse/Davidson line of which we have spoken). Bradley also provides a case where desire and judgment are insufficient to yield volition. He capitalizes on the temporal hiatus between the agent and the "desired end." Suppose today is Monday and I desire my Thursday trip to the grocer, which I judge to be inevitable. Even so, there is no volition present; nor even a present intention. ("Definition of Will (I)," CP. p.500). The reason for bringing 'volition' into the picture, when Anscombe speaks only of "will," is that there is a distinction between the two which if not recognized makes her dilemma difficult to surmount.


If the question of "two knowledges" is going to be successfully addressed, some position, ultimately, will have to be taken with respect to the relation (identity or otherwise) between volition and the action itself. Examining the strange case of attempted telekinesis provides steps in the right direction.

When I try to move a matchbox telekinetically, is it like what I do when I move my arm, but failing? Anscombe says, no; James says, yes. Seeing how these two differ supplies some useful facts in trying to figure out what it is to know our intentional actions.

Suppose I try to get a matchbox to move by force of will, i.e., by merely willing *that* it move. How does this experience differ from that of a paralytic (or someone with an anaesthetized arm) trying to get his arm to move? To answer this question satisfactorily we need to compare more closely attempted telekinesis and attempted bodily movement. Even though they are most alike when they fail; i.e., when they are thwarted different classes of descriptions apply to the resulting failure.

There are occasions where my arm is said to move "involuntarily," but I can think of no occasion where it makes sense to say that a matchbox moved (not 'was moved') against my will or involuntarily. I doubt that even if we could say the matchbox moved "nonvoluntarily" that expression would have same sense as when I say that the movement of my arm, while I was sleeping, was "nonvoluntary." The will to move my arm and willing the matchbox to move appears to involve different senses of 'will'. On this issue, Anscombe, as we shall see, sides with Bradley, who, like Anscombe, distances himself from James. James had maintained, not without appreciable uncertainty, that when I will to move some "distant object" (let's use 'matchbox' for consistency where he uses 'table') the willing involved is the same as when I will to raise my arm.

I will to write, and the act follows. I will to sneeze, and it does not. I will that the distant table slide over the floor towards me; it also does not. My willing representation can no more instigate my sneezing-centre than it can instigate the table to activity. But in both cases it is as true and good willing as it was when I willed to write. (PP. vol. ii. p. 560)

James acknowledges that not everyone is going to be in agreement. He suggests that more is at issue than certain differences in the use meaning of 'will'. There is, he says, a psychological difference, a difference discerned upon realization of the inefficacy of attempted telekinesis. This leads him to distinguish wishing from willing. But according to James if I am able, psychologically, to abstract from consideration the realization of the impossibility of telekinesis then there is no difference in the willing on these two occasions, viz. those occasions where I try to raise my arm and those where I try to move things telekinetically.

Like James, Bradley is uncertain exactly how to proceed, but he adds something to the discussion by introducing a comparison of the failure of telekinesis and the failure of the paralytic who tries to raise his arm.

But, it may be urged, there are cases where the idea does not advance outwards even up to this point. In these cases the idea, remains entirely within itself, and after all there is an actual experience of will. If this does not take place in paralysis, I may be told, it happens often elsewhere. ("The Definition of Will (I)" Collected Papers. Oxford. p. 491)

Anscombe does not consider the comparison with the paralytic, although I suspect it would have benefitted her position. She will, however, side with Bradley in distinguishing the sort of thing that is going on when I try to raise my arm from the situation where I try by force of will to move an external object. For Bradley the will, as it relates to my moving objects by means of it, is required somehow to be connected to a bodily action, and here one would like to say that in the case of telekinesis some bodily action serves as "vehicle" of the intention to move the object. This is what I will consistently take Bradley to mean by a "fixed glance."

For myself, usually, where I will, let us say, a chair to transport itself across the room, I find that I connect this anticipated movement with some bodily act of my own. A fixed glance, an order uttered inwardly or some other slight movement...(ibid)

I used the word "vehicle" before introducing this quote very deliberately. It occurs in section 25 (p. 44) of Anscombe's _Intention_ and it is not to be lightly glossed. In describing the case of the distinterested pumper of poison Anscombe had spoken of a mere grunt as the "vehicle" of his thoughtful indifference. The vehicle here "carries" a thought in contrast to being a tool of communication.

The similarity of Anscombe's position to that of Bradley is easy to miss; probably because we sometimes think of her work as a mere continuation of Wittgenstein's. In a way that would require considerable fleshing out, Wittgenstein's later work can be said to be a repudiation of the Jamesian "idea." A pervasive notion in what I would call the "classical will." We have already seen that Anscombe explicitly raises the issue of the Jamesian idea and is inclined to reject it. Bradley will embrace the "idea," requiring its presence in all "actions." Where Bradley speaks of a "fixed glance," in connection with telekinesis, Anscombe makes use of "stare" in a way which I have suggested renders it a vehicle of thought. Anscombe says,

...for the only sense I can give to willing is that in which I might stare at something and will it to move. (Intention. p. 42)

It is the idea which "prevails" on Bradley's view that determines an action. This explains the "fixed glance" he brings into attempted telekinesis. A fixed glance, according to Bradley "goes together with" prevalence of the idea (CE p. 491). But what about Anscombe, who rejects the Jamesian "idea"? What is the role in relation to Bradley's "fixed glance" of Anscombe's "stare"? And here we are about to reach one of our most important conclusions.

I have discussed elsewhere the importance of the distinction Anscombe makes (never explicitely, but almost) between an intentional action and an actions done "with" an intention. Most likely under the influence of Wittgenstein, Anscombe in rejecting the classical "idea" felt the need to account for its function in other terms. Instead of talking about ideas as such, Anscombe, whenever possible, will speak in terms of descriptions. Being performed "with" an intention (Anscombe) and being performed for the sake of a prevailing "idea" (Bradley) differ mainly in the fact that an intention is not something inner, it corresponds to a description, a description "under which" an action falls. We have seen elsewhere that Anscombe is concerned with "vehicles" of intention. A grunt, she notes elsewhere, may be vehicle for a thought - or for an expression of indifference. Although the textual evidence is very slight, and one could not be sure even if the evidence were considerable, it seems reasonable to believe that the "stare" in Anscombe is a vehicle of my intention to make the matchbox move in the telekinetic case. Being performed "with" an intention, as an act of will, requires a vehicle for that intention; where being such a vehicle is the function of the "stare" in Anscombe's case. For Bradley the "fixed glance" is not an act of will, it is an act of attention and as such lacks the properties we are using to describe the will, most generally conativity (vide "On Active Attention" CE. p. 440). But this is a story that will have to go untold in this paper.

The attention signalled by the stare is intimately connected with the ends of imminent action or what Bradley calls "the positive end which...is pursued in attention." (C. E. p. 413 "On Positive Attention.")Attention may result in an action such as adjusting my hands to catch a ball. In telekinesis attention, however, is as far as the intentional action gets.

Another way of viewing matters is to say that Anscombe's "stare," and Bradley's notion of attention, both imply the same thing: volition. As Bradley says, "Active attention everywhere implies volition" (CE p. 416). 'Volition' as I use it, but not as some others have used it, implies only voluntary action, not action guided by some conscious idea of a desired end. For the latter attention, as Bradley understood, is required. For Bradley "volition" (willing) consists in two stages, one of which is "prevalence" of the idea that gets actualized in completing the action. The fixed gaze is the first stage of telekinesis as an intended action ("Definintion of Will (I)," Collected Papers. p. 486). James, on the other hand, will make prevalence the terminus of willing itself. The best way of thinking of the fixed glance for Bradley is as an "incomplete action."


By "negative" actions I mean actions describable by sentences with negation morphologically realized and occurring syntactically as an adverbial modifier of the main verb indicating action. Consider the following pair of sentences.

A.      I raised my arm?
A'.     I didn't raise my arm

The first sentence, A, is a sentence expressing a 'positive' action; the second, may be used to describe a negative action. Do I know the truth of A' the way I may know the truth of A? That is, can I know A' in some way other than observation? If my knowledge of A is not based on kinaesthetic awareness - and Anscombe leaves this an open question - then how do I know it to be true; and if I do know it to be true, do I ever know the truth of sentences like A' in the same way?

If I know the truth of 'I raised my arm' in some way other than by observation, then even so, I know that my arm went up. But while I may know 'I didn't raise my arm' in some other way than observation, such knowledge does not entitle me to claim to know the truth of 'My arm didn't go up'. What difference here is there between the action and its result? Negative actions, such as this one, are actions only if they are acts of will. They are acts of will on our theory only if they are performed "with" an intention. I would hardly ever wonder why I didn't raise my arm if the right answer were, as it might be in the case of a positive action, 'Oh, no reason'. Because a negative action can only be an act of will, as we understand such acts, when it is performed with an intention, knowing it as an intentional act of ours cannot be the same as the ways in which we may know a positive act, since a positive act may be performed "with" no intention whatsover, even though it were intentional. This presents a difficulty to some who have contributed mightily to the discussion of volition and action. Melden is a case in point. Melden believed that I know of my arm raising actions by other means than sensation. What counts is our ability to make the sentence true which describes the action. Here is what Melden says,

If the doing is simply the raising of one's arm, why not say that one needs no sensation to tell us that one is moving one's arm, no inner signals of any sort, that one is able to tell simply because one is able to *make* the proposition 'I am raising my arm' true'? (Free Action. Routledge & Kegan Paul. 1961. p. 37.

If what Melden suggests is right, then we ought to ask, "Am I able to know the truth of A' because I am able to make it true? The answer would appear to be no. I cannot know that that I didn't move my arm simply by making it true that I didn't didn't move my arm. Melden, almost certainly, felt that his care in emphasizing 'make' would spare him this problem, since I don't *make* my arm not move in my sleep. Aside from the fact that I can't know what I do, if anything, in my sleep, there is the added difficulty that the truth of the proposition: 'I didn't raise my arm', may not refer to an action at all, but merely a circumstance or fact involving no intentional behavior in the least. I would need to know when I can be said to *make* my arm not move, and this presupposes knowledge the *making* was supposed to explain. Another problem is that being able to make the sentence 'I didn't raise my arm' true is insufficient to inform or even 'advise' me that the event is an action of *mine*; a fact paramount in classical theory but barely if ever mentioned by those following Wittgenstein.


According to Anscombe one approach to the question "How do I know my intentional actions" involves saying that the intentional action is carried out in the mind, even if there is no forthcoming empirical result. Recall it is one of Anscombe's objectives, if I am right so far, to suppress the significance of the "idea" which had been overriding in James, Bradley, and many others. Getting from an "idea" (or "mere intention") to an action had always been a problem for the classical theorists. Anscombe calls attention to one of Wittgenstein's few comments on the will in the Tractatus.

Even if what we wish were always to happen, this would only be a grace of fate, for it is not any logical connection between will and the world that would guarantee this, and as for the presumed physical connection, we cannot will *that*. (Tractatus 6.374)

A quarter of a century or more earlier, the classical theorist, G. F. Stout, said something remarkably similar. Only where Wittgenstein had spoken of "grace of fate" Stout had spoken in terms of a "benevolent dispensation of Providence."

The connection between certain modes of consciousness and corresponding movements of the limbs adapted to satisfy our desires is a benevolent dispensation of Providence... ("Voluntary Action" MIND. 1896)

What is the issue here? The main issue is whether the "idea" of classical theory has efficacy. Something James believed quite passionately to be true.

We may then lay it down for certain that every representation of a movement awakens in some degree the actual movement which is its object; and awakens it in a maximum degree whenever it is not kept from so doing by an antagonistic representation simultaneously to the mind. (Principles of Psychology. vol. ii. p. 526)

For the classical theorists, the "awakening" of which James spoke takes many forms. It may be a matter of instinct, or habit, or practical reason. There is no simple formula for characterizing the relation of idea to the world in an intentional action.

Anscombe's objection is to the way the "idea" is made to enter the picture. She wonders out loud how such a thing is possible. What she asks is: What is the "vehicle" of the idea? And what might be the relation of the vehicle and the idea? How could such a relation be guaranteed other than by the "awakening" of which James speaks and being brought into being by either "grace of fate" (Wittgenstein) or "Providence" (Stout)? She insists that there must be such a vehicle. Why, then, is she so insistent? Because, again, she wants to purge the "idea" from intentional action. But she doesn't go so far as to explicitly articulate this concern. She merely says that without it we are left with "a bombination in a vacuum." What, then, is her solution to the problem?

She reminds the reader that elsewhere she has said: "I do what happens." What happens is known by observation, and so the question of two knowledges resurfaces. She, then, illustrates knowing without observation, as when I write with my eyes closed. In such a case, what I write may not be well written; it may be slanted on the page or go off the page, etc. Observation does not so much play the role of letting us know what we are doing but, even with our eyes open, it informs us continually of how we are doing; it directs our actions. It is, as she says, "merely an aid." The point seems to be that I don't actually know my intentional action by observation; observation is simply a guide to action. She concludes by reiterating her claim that if there are two ways of knowing then two things must be known. What are we to conclude? Anscombe appears to be saying that in the writing case I am informed by sensations associated with movement. Such a view is not new. (Wittgenstein's Investigations p. 185).

Although Anscombe continues throughout to be somewhat elusive as to how it is we come to know our intentional actions, it is almost certain that for her it is in the same way as I know without looking the position of my limbs. Without fully embracing the possibility, she appears to accept bodily sensations as offering the relevant clues. Wittgenstein, it will be recalled (PI. p. 185), acknowledged that sensations "advise" us as to how our limbs are situated. He expands on this theme, saying that pain sensations "advise" us as to their location. It would be absurd to say I can also know by observation the location of my pain as I can the position of my limbs, and that, therefore, there are two "knowledges" of a single location of pain. What one ought to say, I believe, is that I can't always know by observation is that the action is mine. In the same way that for Wittgenstein a sensation informs me of the location of a pain, a volition informs me that its consequences in some sense "belong" to me.


Anscombe leaves the dilemma she raises unresolved, considering it mainly as a problem for certain theorists I have described as "classical theorists," that is those who held that action requires ideas to act upon. She maintains that there are two knowledges, one observational of what happens, and another, a nonobservational knowledge of what I do. Two knowledges entail, she argues, two things known. But is this true? The position I have taken here is that there is a middle course, and that my sensations advise me that an action is volitional (that is they inform me that it is I who initiate the action or sustain it), and what I can know by observation is at most what it is that I make happen to me. Such a view can be culled with modifications, drawing from Anscombe's distinction between intentional and willful action, from the classical theorists, such as Bradley who asserted will to be "the self realization of an idea with which the self is identified" (MIND 1902 in C.E. p. 476). We add to the classical theorists by requiring that volition be distinct from the will. Working within Bradley's framework we say that in acts of will what becomes "identified" with the self is the "idea," whereas in the case of volition identification is with an action. Knowledge of what "happened" (to use Anscombe's felicitous expression) implies nothing about the relation of what happens to the agent, whence the problem as conceived by Stout and Wittgenstein which we have already discussed. Knowledge, on the other hand, of what the agent "does" by contrast warrants self attibution. The dilemma which is at the center of section 29 of Intentions is solved by means of this distinction.