Davidson on Knowing Future Successes

Here is Davidson's argument from counterexample:

Davidson also provides this different example, which is presumably intended to ward off Anscombe's claim to the same degree and for the same reasons as the previous:

On the face of things, Davidson's argument is a straightforward appeal to counterexample: Here are two cases where someone is doing something intentionally without knowing they're doing it, and so it cannot in general be true that doing something intentionally requires knowing one is doing it. But I want to argue that if Davidson's argument is to be successful, it cannot be this simple. For it is, on the face of things, deniable that the subjects in the examples counterexample Anscombe's claim, and so there must be some further content to Davidson's argument to show that they do.

For it seems possible to insist that, if someone is providing for their children's welfare intentionally, they know they are doing that. After all, a project of that sort must then be underway, and it would be strange to think that the responsible subject is not aware of this fact. Or (to try the reverse approach with the other example) we could say that if the carbon copier doesn't know that they are making ten legible copies, they are not doing that intentionally. After all, if it is uncertain that there will be success in making the ten copies, then in this respect the case is like one of picking the winning number in a lottery, of which we would not say that a player is winning intentionally.

The previous paragraph is not meant to show that Davidson has no case against Anscombe's claim. It is merely intended to show that his case can't be a simple appeal to how we (‘certainly’) should describe the above pair of imagined subjects.Footnote ¹⁴ So given that Davidson's examples are not pre-theoretically clear cases of doing something intentionally but unwittingly (as, by contrast, a Gettier case might be thought to be a pre-theoretically clear case of justifiably truly believing something without knowing it),Footnote ¹⁵ just what is it in these examples that is supposed to tempt us to think that their subjects are doing things intentionally but unknowingly?

Roughly stated, the thought behind Davidson's examples seems to be this: Doing something like providing for the welfare of one's children through a will, or making ten copies by pressing through ten pieces of carbon paper, requires that things take a course which stretches beyond the knowledge that a subject necessarily has at the time of doing it. But it still seems right that a subject can be doing either of these things intentionally. And this is why we should think that doing something intentionally doesn't require doing it knowingly. But appealing as it might be, this formulation of the argument is too rough, since it is not clear in what sense providing for one's children, or making ten copies, is supposed to ‘stretch beyond’ what such a subject necessarily knows when they are doing it intentionally.

To see a way of capturing this intuitive sense of stretching beyond, consider the difference between the doings that support Davidsonian challenges and those that do not. The doings that support Davidsonian challenges are innumerable, and include the following:

• making ten copies of a piece of paper

• providing for the future welfare of one's children

• getting to the moon

• winning a race

• outliving one's arch enemy

For each of these, it seems possible to argue, in Davidson's style, that someone can be doing it intentionally, without knowing whether in fact they are doing it. If someone is, say, winning a race while blindfolded, it will be tempting to say that this fact stretches beyond their cognitive reach, although they are winning it intentionally. And someone who is outliving their enemy or getting to the moon may not know whether they are, although if they have sworn to do it and are doing their best, it is arguable that they can be doing it intentionally.

Let's now consider the doings that don't support Davidsonian challenges. Again we have innumerable examples, including those mentioned at the start of this essay and more:

• pushing a stroller

• cutting a strawberry in half

• begging for an extended deadline

• building a helicopter

• walking

• booing

We cannot make a Davidsonian challenge by supposing that someone is cutting a strawberry in half intentionally, while unsure about whether they are cutting it in half. And the reason for this seems to be that if they are doing it intentionally, there is no clear way of making sense of the suggestion that this episode of cutting ‘stretches beyond’ their cognitive reach, as outliving, winning, and making ten copies might. Things seem to be the same with the other cases in the list: If someone is intentionally begging for an extended deadline, then how can they fail to take in the fact that they are begging for it? The same goes for walking and booing.

So what is the sense in which the first few doings, and not the second few, can stretch beyond a subject's cognitive reach, thereby making it seem possible to do them intentionally but unknowingly? Someone might suggest that the difference has to do with the difficulty or complexity of the tasks, and the corresponding difficulty of being certain that things are going according to plan. But this thought is a dead end, since building a helicopter and getting to the moon can both be difficult and complex undertakings, where a participating subject may not know how things are going. It still seems that only the latter case allows us to pose a Davidsonian challenge.Footnote ¹⁶

But here is a truth about the entries in the first list: If a subject is doing one of these things, then it must be that they have not yet done it, although it must also be that in the future they will have done it. So if someone is winning a race, they can't yet have won the race, but it must be true that in the future they will end up having won it. And if someone's getting (as opposed to going or travelling) to the moon, they can't yet be at the moon, but they must be such that they will end up having gotten there in the future. Equally we can't accurately say that someone is outliving their arch enemy if they have already outlived them, of if they will never end up having outlived them.Footnote ¹⁷

I want to call the doings that fall under this truth ‘achievements’. In using this word for these doings, I partially agree with a tradition that runs from RyleFootnote ¹⁸ and VendlerFootnote ¹⁹ through MourelatosFootnote ²⁰ to ThompsonFootnote ²¹ and Hornsby.Footnote ²² For we all count (at least roughly) the same doings as achievements, particularly including things like these: winning a race, getting to the other side of the street, finding a shoe, and dying. But the agreement is only partial, since these authors do not always overtly endorse my criterion for achievements, and tend to prefer defining them as somehow ‘punctate’Footnote ²³ or ‘non-durative’.Footnote ²⁴

Below I want to briefly explore the commitments one would have to take on to think of achievements as durationless. (This is not beside the point of this essay, since it will turn out that there are several respects in which our further commitments about the nature of achievements will determine our stance on the emerging Davidsonian argument.) We may take as our starting point Ryle's brief discussion of winning:

Generalising Ryle's claim, we arrive at the claim that achievements, like winning, are durationless. But there are two competing ways of spelling out Ryle's thought. For we may say either that achievements are durationless because, in fact, no one is ever doing one of them (though lots of people and things have done one or more of them), or, instead, that although a subject can be doing an achievement, this does not go on for any stretch of time, but rather happens instantaneously. It seems to me that both options come with their own small difficulties:

Let's take the first line, and say that nothing is ever doing an achievement. This arguably gives a distorted picture of achievements. This is noted in passing by Hornsby, who says it can be fine, in ordinary language, to describe someone as ‘finding the book while she is searching for it’,Footnote ²⁶ even though finding seems like an achievement. Mourelatos has similarly noted that ‘[o]ne can easily cite or compose well-formed sentences in which any of the verbs [of] achievement […] appear in the progressive’.Footnote ²⁷ It is true that there are apparent ways of defending the present suggestion from such linguistic observations – maybe ordinary language is wrong, or maybe it needs a subtler interpretation. But I don't think any such defence has been pursued in much detail.

Let's then try the second line, and say that something can be doing an achievement, but only in a temporally non-extended sort of way. We immediately face two further options. We may take Hornsby's and Mourelatos's sentences at face value, and say that someone can be doing an achievement before the instant when they have gotten it done, or we may declare that someone can only be doing an achievement at the instant when they have gotten it done.

On the first view it becomes harder to see how achievements could remain durationless. For if right now I am finding a book, and if soon I will end up having found it, then how could my finding it fail to be the sort of thing that can go on for a stretch of time which ends when I have found it? Mourelatos seems to answer that achievements are performed ‘during’ stretches of time, but not ‘throughout’ such stretches.Footnote ²⁸ Perhaps this means that although someone can be finding something at many instants during a time before they have found it, they can't be finding it for any stretches of time. Mourelatos's distinction may be questioned: Why shouldn't this multitude of instantaneous moments of finding add up to an episode or stretch of finding? This question seems difficult to me, but maybe it has an answer.

On the second view, a subject starts and finishes doing an achievement in a single instant. This view does give clear sense to the suggestion that achievements are durationless, but has its own sort of oddness. For in the particular case it would have to come to this: Someone can be winning a race only at a point-like moment, with no temporal extension, right in between the time when they're in the race and the time when they have finished it. This again introduces an issue of how to reconcile the view with ordinary claims about what people are doing: Is saying that someone is winning before they have won simply to speak falsely, or is it really a tacit way of referring to the fact that they will win later? As before, I am not arguing that the answer to these questions must be negative, but just suggesting that we don't have a ready answer to either.

Our own view of achievements is, on the face of it, neutral on the question of whether achievements have durations, and also neutral on each of the above three conceptions achievements as durationless. In fact, it seems that each attempt to say that achievements are durationless can be read as an attempt to explain why, in the case of achievements, future success is guaranteed by present engagement. Why can't I be winning a race, and later fail to have won it? Perhaps because, strictly speaking, I can never be winning a race (so that facts about winning can merely be retrospective), or perhaps because winning is never going on throughout a stretch of time (as on Mourelatos's somewhat elusive conception), or perhaps because winning only happens at the instant where I become the winner (so that, as soon I have started winning, I have also finished doing it).

This view of achievements gives us a ready way of making sense of the claim that, if someone is making ten carbon copies of a piece of paper, or providing for their children's future welfare through a will, this undertaking can stretch beyond their cognitive reach. For both these undertakings seem to be achievements: I cannot now be providing for the future welfare of my children, unless in the future their welfare ends up provided for. And I cannot now be making ten copies of a piece of paper, unless in the future ten copies end up made. But it is precisely their standing as achievements which seems to put them outside the cognitive reach that a subject necessarily has of what they are doing intentionally: For if I'm to know I'm doing one of these things, I must presumably know that in the future I will have the requisite kind of success. But it seems possible for me to do things intentionally without having such predictive knowledge of such a future success.

If it is right to think that Davidson's appeals to counterexample were really tacit appeals to this pair of concerns, we may now formulate this sharper and more abstract version of his argument:

• Knowing Achievements: For some of the things a subject can be doing intentionally (including providing for one's children's welfare through a will, or making ten carbon copies of a piece of writing) knowing that someone is doing it requires knowing that, in the future, they will end up having done it

• No Prediction: For all of the things a subject can be doing intentionally, someone can be doing it intentionally without knowing that, in the future, they will end up having done it

• (C) No Knowledge: For some of the things a subject can be doing intentionally, someone can be doing it intentionally without knowing that they are doing it

This formulation of the argument seems to preserve the intuitive swaying power of Davidson's original pair of objections, while also clarifying where the force of these objections is supposed to come from. But on reflection, it will emerge that both premises in this argument are more fragile and uncertain than Davidson has made apparent.

A very straightforward way of responding to the argument denies Knowing Achievements. A natural elaboration of this denial goes on to claim that achievements – those undertakings where knowing present engagement requires knowing future success – can't be performed intentionally. This claim might be grounded in the simple suggestion, mentioned above, that nothing is ever doing an achievement – these being merely retrospective facts about a subject's past successes. Or it may be grounded in the different suggestion that achievements fall outside the scope of what a subject can be doing intentionally.

As was mentioned before, the first idea seems to generate some need to say why achievements are sometimes ascribed in the present progressive. But this is arguably a mild concern. In any case, the second idea does not generate such a need. On this idea, saying that someone is winning may be strictly true, although no one can be winning intentionally. In fact, it seems to me that two of the aforementioned three conceptions of achievements can ground this idea. In the next two paragraphs I will explain how.

Suppose that Thompson and Hornsby are right, so that achievements are durationless. Can't we then deny the first premise of the Davidsonian argument by saying that doing something intentionally, properly speaking, is doing something that takes time? After all, it is arguably only where there is temporal extension that a desire may cause an agent to do something, in the sense of ‘cause’, whatever it is, that is proper to intentional agency. If we think that the doings that must be predicted successful to be known are those that don't take time, and that those doings can't be performed intentionally, then we have in this the makings of a satisfying way of grounding our rejection of the Davidsonian argument.

Or suppose, instead, that something like Mourelatos's conception is correct: There are several different times at which someone can be truly said to be winning a race, but still winning is not the sort of thing that is done ‘throughout’ a stretch of time, in the sense that, say, running is. Can't we now declare that doing something intentionally is doing it throughout a stretch of time? Though this proposal is structurally similar to the previous, it is admittedly difficult to tell just how good it is. For we have no clear hold of Mourelatos's conception of ‘throughout’. But assuming that Mourelatos is not confused, so that there is a clear way of working out the throughout/during-distinction, there does not seem to be a prima facie difficulty about stipulating the former to be the proper province of intentional agency.

If someone should object that either response leads to a too restrictive conception of what an agent can do intentionally, it is worth noting that, wherever someone might be tempted to say that someone is intentionally doing an achievement, it seems very possible to say, instead, that they are intentionally doing something which, if things go well, might result in them having done an achievement. Hence instead of saying someone is intentionally winning a race or securing the favour of their uncle, we may say that they are intentionally participating in the race – perhaps very successfully or ferociously – and may therefore end up having won it, or that they are intentionally blandishing their uncle, and therefore may well end up having secured the uncle's favour. This way of describing these proceedings is really no less natural than that which it would replace, and is arguably more natural than it.

In sum each of the three considered conceptions of achievements (the one on which nothing is ever doing an achievement, the one on which these doings are durationless, and the one on which they do not go on ‘throughout’ stretches of time) seems to give us a way of denying the first premise of the Davidsonian argument. The first one makes short work of the idea that someone can be doing an achievement intentionally, because it simply denies that anything can be doing an achievement. The second and third require more elaboration, but I have tried to describe how they might work. Let's now move on to a different type of response to the Davidsonian argument, which denies Non-prediction.

On a natural elaboration of this response, someone could say that where someone is intentionally doing an achievement, they do have a predictive, true belief about future success in so doing. (This might be true even if, as a general matter, a subject who is doing something intentionally need not know that success will come.) Hence if someone is intentionally making ten carbon copies, where this is construed as entailing that they will end up having made ten copies in the future, then this subject knows this fact about the future. Is there really anything incoherent or problematic about such a subject?

Of course there are arguments that threaten any knowledge of the future, but Davidson's argument is surely not supposed to draw on something that radical. Someone might then try to support No Prediction by saying that predictive knowledge of a future success somehow makes it impossible to intend such a future success. As Setiya has noted, there is something problematic about the idea that a subject intends to do something which this subject thinks they will end up having done come what may, but this does not mean that, more generally, there is something problematic about a subject who intends to do something which this subject thinks they will end up having done sans phrase.Footnote ²⁹

I do not know of another way of undermining the suggestion that, where a subject is intentionally doing an achievement, this subject knows that they will end up having done it in the future. And so I don't see why No Prediction need be accepted.