Intending Recalcitrant Social Ends

Abstract

One can intend the actions of others, even when one believes such actions are not under one’s control. Call the objects of intentions “ends”; the ends that consist (partly or wholly) of other people’s actions “social”; and the ends that consist of things one believes one cannot control “recalcitrant”. The thesis, then, is that one can intend recalcitrant social ends. I present a positive argument in favor of this idea, and then argue against some purported conditions on the possible content of intentions that would rule out the possibility of intending such ends. The positive argument is the following: intentions are realized by a certain cluster of dispositions. One can come to be so disposed towards recalcitrant social ends. So one can come to be disposed toward such ends in the manner that realizes an intention. So one can intend recalcitrant social ends. The conditions against which I argue are known in the literature as the Own Action Condition, the Control Condition, and the Settle Condition. I present cases where, intuitively, agents genuinely intend ends that would be ruled out by these conditions.

1 The Thesis

One can intend the actions of others, even when one believes such actions are not under one’s control. For example, I currently intend that my friends and I get together for beers on Friday, even though I believe that whether we do so is not—in any interesting sense—under my control. I can invite them, I can try to convince them to come, but they may have better things to do and decide not to join. Call the objects of intentions “ends”; the ends that consist (partly or wholly) of other people’s actions “social”; and the ends that consist of things one believes one cannot control “recalcitrant”. The thesis, then, is that one can intend recalcitrant social ends.¹

I think most philosophers of action believe that one cannot intend such things. The reasons they have for thinking this are various, but the common idea goes something like this: an agent can only intend what she believes is “up to her”, where this idea is cashed out in different manners.

For example, it has been proposed that the objects of intentions are somehow limited to the agent’s own actions; or, if not to her own actions, at least to what she believes she can control; or, if not to what she believes she can control, at least to what she believes her so intending somehow settles.

In saying this, I am following a list of putative conditions on the possible contents of intentions that I take from Michael Bratman (1999). The conditions (roughly) are the following:

• 1. Own Action Condition: one can only intend one’s own actions.

• 2. Control Condition: one can only intend what one believes one can control.

• 3. Settle Condition: one can only intend what one believes one’s so intending somehow settles.

None of these conditions serve to rule out recalcitrant social ends as possible objects of intention. The first condition would rule out such ends if it were true, but it isn’t. Intentions can, and commonly do, take as contents states of affairs that consist of the actions of others, or that consist of no actions at all. The second and third conditions, in contrast, are perhaps true under a certain way of interpreting them. But the way of understanding them that would render them true does not rule out recalcitrant social ends as possible objects of intention.

I should qualify the thesis from the start: my concern in this paper is with the mental state that plays the functional role that prominent philosophers of action have attributed to intentions. My claim is that that mental state can take as its objects recalcitrant social ends. What I say about this state may not map well onto the way English speakers apply the word ‘intention.’ So my claim here is not that the state—if there is such a thing—that is picked out by English-speakers’ linguistic propensities can take ends like these as its objects. My claim is about a mental state that plays a specific functional role—the role in virtue of which that state has been found to be explanatorily powerful and thus deserving of a central place in our theory of mind and action.

I will begin in the following section by presenting a positive argument in favor of the idea that one can intend recalcitrant social ends. I will then argue, in Sects. 3, 4 and 5, against each of the three conditions mentioned above.²

2 The Positive Argument

In this section I will present a positive argument in favor of the idea that one can intend recalcitrant social ends. The argument goes like this:

• 1. Intentions—I assume—are constituted by their functional role.

• 2. So, if a certain state of mind plays the functional role of an intention, that state realizes an intention.

• 3. There is a specific cluster of dispositions such that, if one is disposed in such a way towards an end e, then one is in a state of mind that plays the functional role of an intention that e.

• 4. So that disposition realizes an intention that e.

• 5. One can come to be so disposed towards recalcitrant social ends.

• 6. So one can come to be disposed towards recalcitrant social ends in a way that realizes an intention towards them.

• 7. So one can intend recalcitrant social ends.

I will now consider each of these premises in turn.

Premise 1 is an assumption I will not try to defend. I assume a broadly speaking functionalist theory of mind according to which mental states are constituted by their functional roles.

Premise 2 is a central tenet of functionalism: if a state of mind plays the functional role of an intention, then that state is an intention—that is, it realizes one—for intentions are what intentions do.

Premise 3 is the core of the argument. I cannot defend it in full detail, but I think the main idea can be conveyed in broad strokes:

Turn first to the functional role of intentions. What is it exactly that intentions do? Obviously, there is discrepancy on this issue. Nevertheless, I think it is possible to extract a core functional profile from what prominent theorists tell us about the role intentions play. Here are two representative views on this issue:

Michael Bratman (1987) tells us that intentions constitute a kind of commitment to a goal. He calls this a “practical commitment”, and he says that it involves two different dimensions: one “volitional” and one “reasoning centered”. The volitional dimension is related to the way in which intention controls action. It is characteristic of intention that, if one intends to φ at time t, the intention persists until t, and one is relevantly able to φ at t, then one φs at t. (If one is not relevantly able to φ at t, one tries to φ at t.) Intention is in this respect different from desire. After all, it is perfectly normal to desire to do something, be in all relevant respects able to do it, and yet not do it (or even try to do it). Intention, Bratman says, is “conduct-controlling”, whereas desire is merely a “potential influencer” of action (1987 p. 16).

The reasoning-centered dimension is related to the way in which intention frames practical deliberation. There are at least three distinctive aspects of this dimension. Intention, according to Bratman:

• Settles practical deliberation (at least in the absence of reasons to reconsider);

• Poses means-end problems, thereby prompting the agent to figure out, and intend, means believed to be appropriate to her ends;

• Constrains the formation of new intentions by filtering out as possible further ends any state of affairs that the agent believes to be inconsistent with her intended ends.

Intention is different from desires and other motivational states in each of these respects. For example, one can desire to φ and not see the question whether to φ as settled. In fact, one can desire to φ and be settled on the goal of not φing. Because of this, one can adopt further goals that one sees as inconsistent with φing, and one need not be concerned with figuring out and intending to take appropriate means to φ. (1987 pp. 16–18)

Moreover, according to Bratman, it is in virtue of these features, together with its conduct-controlling nature, that intentions can (and desires and/or beliefs cannot) play two further distinctive roles: first, they can extend the influence of an agent’s deliberation over time, and, second, they can support complex forms of coordinated, temporally and socially extended activities.³

This is in line with what Alfred Mele (1989, 1992) says about the role of intention. According to him, the “functional core” of intention is that of a state which “initiates and motivationally sustains intentional action; guides and monitors behavior; coordinates one’s activities, including one’s interaction with others; and can both prompt and suitably terminate practical reasoning”. (1989 p. 27)

It is in virtue of these features that Mele takes intention to be different from desires or any other motivational attitude (alone or in combination with beliefs), and in virtue of which he thinks we need to incorporate intentions as genuine attitudes into our theory of the mind and its relation to action.⁴

I think there is sufficient consensus in the literature that intentions do, indeed, play these roles. Succinctly (and roughly) put, then, let me say that the functional role constitutive of intentions is the following:

• Functional Role of Intentions: intentions are attitudes that motivate and control an agent’s behavior, by way of settling her deliberation on the pursuit of a goal, guiding and monitoring her progress towards it, and framing her further deliberation in the direction of believed consistency among intended ends, and believed coherence among intentions for ends and intentions for appropriate means. In doing so, they help extend the influence of the agent’s deliberation over time, and they support complex forms of coordinated, temporally and socially extended activities.

Although there is consensus that intentions do at least this much, some theorists hold that intentions do something more than this.

For example, theorists who think that an intention that e just is, or in any case metaphysically implies, a belief that e, have suggested that a further functional role of intentions is that they contribute their propositional contents as fixed points for further practical deliberation (so that, if an agent intends that e, she further deliberates about what to do on the assumption that e).⁵

It is true that intentions do this to a certain extent. But this is not an essential role of intentions. That is, it is true that, frequently, when you intend that e, you deliberate on the assumption that e (i.e. you evaluate further possible courses of action conditional on e). But this is not necessarily, or even generally speaking, the case. It is not even necessarily the case in those situations where you do believe that e.⁶ Consider: you intend to walk to work before it starts raining. The clouds still seem far away, so you believe you will be able to do it. However, you are not 100% sure of this (let’s say you are 99% sure). Conditional on your walking all the way to work before it starts raining, not carrying an umbrella is the dominant strategy, since it is a bit uncomfortable to carry it around. Suppose, however, that getting wet would be terrible. In such a scenario, you would not deliberate whether to carry it or not on the basis of your belief that you will manage to walk all the way before it starts raining. Instead, you would deliberate on the basis of the probabilities you assign to the relevant contingencies. Relative to such probabilities, the reasonable thing to do (we can assume) is to carry the umbrella. So you would decide take the umbrella. This kind of phenomenon permeates all of our lives, and it shows that intending that e does not necessarily, or even in general, involve a disposition to make further planning on the assumption that e. In fact, it shows that not even belief that e necessarily involves such a disposition.

However, if intention were, or metaphysically implied, a corresponding belief (that is, a belief with the same propositional content), it could involve such a disposition to the extent that belief does. So the more central question is whether intention metaphysically implies a corresponding belief. I think it does not. Most theorists agree that it does not. But the issue remains contentious. Arguments to the effect that it does not have been presented by Davidson (2001), Brand (1984), Bratman (1987), Adams (1986), Ludwig (1992, 1995), Mele (1989, 1992), McCann (1986, 1991), Holton (2009), and others. I take these arguments to be conclusive.

At the same time, my worry with the thesis that intention implies a corresponding belief is perhaps more basic, and it is that the exact same considerations that have led philosophers of action to incorporate intentions as genuine attitudes in our theories of mind and action would—under the stipulation that intentions do imply corresponding beliefs—lead us to incorporate an attitude that is just like intention so conceived, except in that it does not imply a corresponding belief. This is because the same settled, deliberation-framing and goal-directed behaviour that seems unaccountable for in terms of mere beliefs and desires (or other combinations of attitudes) can obtain relative to a goal that the agent does not believe will obtain. (This will typically be the case when the agent assigns a value to the obtainment of the goal that is very high relative to the costs of pursuing it—so much so that the expected utility of adopting and pursuing the goal will be greater than that of not doing so, despite the relatively low credence that it will obtain.) But, if this is so, then the issue whether intention implies belief starts to seem purely linguistic. Call this other attitude that is just like intention, except in that it does not imply belief, a “practical commitment” (to use Bratman’s phrase) to a goal or end. Once we have this item available in our theoretical toolbox, we can simply note that it is sometimes accompanied by a corresponding belief. When it is, we can go ahead and call it an ‘intention.’ I would be happy to oblige to this stipulation. But, as far as I can see, it serves a merely linguistic purpose.⁷

From now on, then, I assume that intention does not imply a corresponding belief. If the reader would like to reserve the word ‘intention’ for an attitude that does imply a corresponding belief, then I what I go on to say applies not to intention but to an attitude that is goal-setting, conduct-controlling, action-guiding, and deliberation-framing in the ways described before. This attitude is sometimes, though not always, accompanied by a corresponding belief.

I now turn now to the dispositional profile that, I claim, can play the functional role of an intention, and so realize an intention.

Again, I think it is possible to identify a core dispositional profile from what prominent theorist tell us about the dispositions characteristic of intention. I won’t try to specify such a profile in full detail. Roughly, though, the idea is that when an agent A intends an end e, then (absent reasons to reconsider):

• End Stability (D1): e is for A a settled, cross-temporally and counter-factually stable goal. e’s being a goal of A involves, roughly and inter alia, a disposition to monitor progress towards e by comparisons between A’s representations of e and actual states of the world at different times, through feedback mechanisms, and dispositions to adjust behaviour in response to such comparisons.⁸e’s being a settled and stable goal of A means, roughly, her being disposed so that, absent new reasons to reconsider, A will neither keep deliberating on whether to pursue e, nor abandon it; she will stick to it as a settled object of pursuit. Call this a disposition of “end stability”.⁹

• End Consistency (D2): A is disposed in practical deliberation to filter out as a possible further end any state of affairs s she believes to be inconsistent with e (in the sense that, were it to be the case that s, then because of that it would not be the case that e). Call this a disposition of “end consistency”;

• Means-End Coherence (D3): A is disposed to figure out, to intend to take, and (if relevantly able) to actually take, appropriate means m to e (among them are the means “implied” mi, such that A believes that, were she not to intend mi, then because of that it would not be the case that e).¹⁰ Call this a disposition of “means-end coherence”.

Call the dispositional cluster composed of D1, D2 and D3, the “i-disposition.”¹¹ My suggestion now is that being so disposed (“i-disposed”) towards an end e would play the functional role of an intention that e, and so realize an intention that e. What follows is a simple sketch of why this would be so:

First, the motivating and conduct-controlling role of intention is secured mainly by D3, the disposition to figure out, intend to take, and (if relevantly able) actually take, appropriate means m to e.¹²

Second, the guiding role of intention is secured, first, by e serving as a goal of A (that is, as a target relative to which she monitors progress and adjusts her behaviour so as to track it) and also by the further dispositions to deliberate in the direction of end consistency and means-end coherence.¹³

Third, the settled and stable role of intention in motivating action is secured by A’s disposition not to keep on deliberating whether to pursue e, and not to abandon e as a goal, in the absence of reasons to reconsider.

Finally, the two further roles of extending the influence of deliberation over time and of supporting complex forms of coordination is secured by the features already acquired. An agent who is in such a state of mind is in an appropriately settled and stable, action-motivating, action-controlling, and action-guiding-state that works in part by way of framing her deliberation in the direction of end stability, end consistency and means-end coherence. If Bratman’s arguments in (1987) are along the right lines, then this state allows an agent’s deliberation at a time to influence her action at a later time, and it supports the complex forms of coordination, both intra and inter-personal, characteristic of temporally and socially extended human agency.

What follows from this is that, if the functional role identified previously is, indeed, the functional role constitutive of intentions, then being i-disposed towards an end e is being in a state of mind that plays the functional role of an intention that e.

Notice, moreover, that the i-disposition can easily explain why it would be the case that, often, when one intends e, one believes that e, and is disposed to further deliberate on the assumption that e. Oftentimes (though certainly not always or even generally) if one does not believe one can bring about the obtainment of a state of affairs e, it makes no sense to adopt it as an end and pursue it, because it would be a mere waste of time and resources to do so. Of course, this can change when the value one assigns to e is so great relative to the costs of pursuing it, that it would make sense to pursue it even when one does not believe one can secure its obtainment. Oftentimes, then, when one intends an end e, one believes that one can bring it about that e. Given the stability and conduct controlling nature of intentions, then, when one intends e, one will frequently be in a position to believe that e: this is both because one will believe that one can bring it about, and because one will believe that one’s intention will be stable and will conduct and guide one’s behavior towards it. (When one is not in such a position, one will frequently at least be in a position to believe that one will try to bring it about that e.) Plausibly, believing that e involves a defeasible disposition to deliberate on the assumption that e. Frequently, then, when one intends that e, one will be disposed to make further planning on the assumption that e (or at least that one will try to bring it about that e). So the i-disposition can not only account for the functional roles essential to intention, it can also account for those roles that are contingently, albeit frequently, tied to it, like that of contributing their propositional contents as fixed points for further practical deliberation.

This concludes my discussion of premise 3, which is the core of the argument.

Premise 4 follows straightforwardly from premises 2 and 3. According to 2, if a certain state of mind plays the functional role of an intention, that state realizes an intention. According to 3, the i-disposition plays the functional role of an intention. It follows from this that the i-disposition realizes an intention. In other words, being i-disposed towards an end e is being in a state of mind that realizes an intention that e.

Premise 5 says that one can come to be i-disposed towards recalcitrant social ends. This, I take it, is uncontroversial. Consider: I said before that I intend that my friends and I get together for beers on Friday, even though I believe that whether we do so is not—in any interesting sense—under my control. Most people balk at this suggestion. However, it is uncontroversial that I can come to be i-disposed towards the goal that consists of my friends and I getting together for beers on Friday, even though I do not believe it is under my control. Call this state of affairs “BEERS”. It is uncontroversial that:

• D1: BEERS can be for me a settled, cross-temporally and counter-factually stable goal. (That is, I can represent this goal-state, monitor progress towards it, adjust behaviour so as to track it, and, absent reasons to reconsider, neither keep deliberating whether to pursue it, nor abandon it as an object of pursuit);

• D2: I can be disposed to filter out as possible further ends any state of affairs I believe to be inconsistent with BEERS. (So, for example, I can be disposed not to adopt the goal of reminding one of my friends that he owes me money, given that I believe that, were I to do so, then because of that he would not join us for beers.)

• D3: I can be disposed to figure out, to intend, and then (if relevantly able) to take, the means I believe to be appropriate to BEERS. (So, for example, I can be disposed such that, if I believe that in order for BEERS I must send some emails or make some calls, then I will be disposed to intend to do so, and (if relevantly able) to actually do so, etc.)

This illustrates that one can come to be i-disposed towards recalcitrant social ends. This is premise 5.

If the argument so far is correct, this cluster of dispositions realizes an intention. I can be so disposed towards BEERS. So I can intend that BEERS.

Put more generally, then, one can come to be disposed towards recalcitrant social ends in a way that realizes an intention towards them. This is premise 6.

The conclusion, then, is that one can intend recalcitrant social ends.

This is the positive argument in favor of the idea that one can intend such ends. It is presented in broad strokes. For one thing, I didn’t argue that the functional role identified is indeed the role constitutive of intentions. I simply appealed to the authority of what I take to be our best theories on this matter. For another, I’ve only gestured at how the kinds of dispositions cited could play that role. So more can be said in support of it. But for my purposes here, it will have to do. I will turn now to a critical discussion of the three purported conditions on the possible objects of intention.

3 Own Action Condition

We often intend ends that consist of our own actions. But we often intend ends that do not. A driver heading towards you, for example, might intend that you realize that your lights are off, and he might also intend that you recognize his intention (Grice 1957); a father might intend that his son clean his room (Bratman 1993); a commanding officer might intend that his troops do something (Velleman 1997); Judy might intend not to go out tonight (Harman 1976); and a dinner host might intend that Bud and Pearl sit next to each other, or that the spoons be in the teacups (Vermazen 1993). All of these, and infinitely many others, are possible objects of intentions.

Some people deny this is possible. They think intentions can only take the agent’s own actions as their object. They accept what Michael Bratman has called the “Own Action Condition”.¹⁴ Because they accept this condition, they are forced to re-describe the intentions mentioned above: they would say, for example, that the father does not really intend that his son clean his room, but only to do something to bring it about that he cleans his room, and so on. Bruce Vermazen called this the “piggy-back” thesis.¹⁵ It is the thesis that:

apparent expressions of non-act intentions are deceptive, and that they should be treated as elliptical expressions of the agent’s intention to do something connected in a direct and simple way with what that apparently intended proposition is about… non-act intentions must always ride piggy-back on intentions to act. (1993 p. 253)

If this thesis has any content to it, then to intend to bring it about that a certain state of affairs obtains is to intend to do something that you believe would somehow cause, causally promote, or constitute that state of affairs. If this is so, then the piggy-back thesis is clearly false.

To see this, consider a case involving authority over the behaviour of other people. Many philosophers think that there is nothing problematic about intending the behaviour of others when one has authority or control over them.¹⁶ Suppose this is one of such cases: a General from the army—let’s say—intends today that his troops march south tomorrow, and is certain that his orders will be obeyed. The proponent of the piggy-back thesis would say that the General does not really intend that the troops march south, but only to do something to bring it about that they do. But this is not plausible. Suppose that this morning the General issues the order that his troops are to march south tomorrow. Since he is certain that his orders will be obeyed, after he has issued this order, he believes there is nothing else he must do to bring this about. He has issued an order, and he is confident it will be obeyed. Because of this, later in the day he does not intend to do anything that will somehow result in, or promote, or constitute, his troops’ marching south. So he does not in the afternoon intend to do something to bring it about that his troops march south. Still, he intends then, in the afternoon, that his troops march south tomorrow, and he will retain such an intention in the usual manner, until the troops have done so and he comes to believe that his intention has been fulfilled.

It seems, then, that the proponent of the piggy back thesis would be forced to say that the General ceases to intend that the troops march south as soon as he issues his order. For, having done what he thinks will bring it about that they march south, he will, in the natural way, cease to intend to do something to bring this about. But it would be absurd to think that he ceases to intend that the troops march south as soon as he issues the order. Suppose that he issued the order in a ciphered manner to his Lieutenant, who has some difficulty deciphering it. He thinks it says that the troops should march south, but he wants to confirm that this is really what the General intends. Should he ask the General then, in the afternoon, if he intends that the troops march south, then the General would answer, then, that that is exactly what he intends.

That the General still intends this in the afternoon is made clear—in fact, it is made true—by how he is disposed to think, act and react in relation to such a possible state of affairs. Earlier I identified a complex dispositional state that, I claimed, realizes an intention. I should simply note that the General is disposed precisely in this way towards the state of affairs that consists of his troops marching south.

That the General still intends this is also made clear—and, some would argue, true—by the fact that, even after he has abandoned his intention to do something to bring it about that his troops march south, he is still subject to rational requirements to which he would not be subject if he merely desired (wished, hoped, preferred, etc.) that his troops march south—e.g., if the General believes that cutting provisions would prevent the troops from marching south, then it would be irrational for him to intend to cut provisions. But he would not be irrational to intend this if, having abandoned his intention to do something to bring it about that his troops march south, he now merely desired (hoped, wished, preferred, etc.) that they do so.

So the General still intends that his troops march south, and it would be rather incredible to suggest otherwise. Therefore, the proponent of the piggy-back thesis would have to argue that, despite appearances, the General still intends to do something that will bring this about.

I said that the General remains disposed to think, act, and react in characteristic ways in relation to the state of affairs that consists of his troops marching south. This set of dispositions realize his intending that the troops march south. The proponent of the piggy-back thesis may latch on to this idea, and go on to insist that what the General intends to do, so as to bring it about that his troops march south, corresponds to this set of dispositions: he intends not to reconsider or abandon this goal; to monitor progress towards it; to not adopt goals he believes to be inconsistent with it; to take appropriate means if he comes to believe that such means are needed, etc.

One initial problem with this suggestion is that it points to intentions that are not directed at the agent’s own actions: two of them, for instance, are directed at his not doing something, and there are good reasons to doubt that these can be analyzed in terms of intentions to act.¹⁷ But leave this worry aside: even then, we have no reason to believe the suggestion in the first place.

Obviously, the reason to believe this cannot be simply that the agent is disposed in all these ways, since one can be disposed to do plenty of things one does not intend to do. Likewise, the reason to believe this cannot be a previous commitment to the Own Action Condition or the piggy-back thesis, since it is the truth of these claims that is currently at issue. Other than these, though, I do not know what reason there could be to believe that he intends such things. On the contrary, I think there are compelling reasons to deny it:

First, suppose that the General is confident that he will not reconsider or abandon his goal. Why, then, would he intend (as opposed to merely being disposed) not to abandon it? There is no denying that, sometimes, when one has the goal e, one may also form the intention not to abandon e as a goal. These are typically cases where one is doubtful of one’s own resolve, because one expects that there will be some kind of temptation along the way that will make one falter.¹⁸ For example, I may have the goal of not smoking and on top of that the intention not to abandon this goal. But this is not normally the case. In fact, it is normally not the case when one has no reason to fear that one will abandon one’s goal—which we can safely assume is the case of the General.

Second, the General need not be concerned with his own mind in the way that this suggestion would have him be. The General is concerned with his troops marching south, not with the management of his own mind. The focus of his “practical attention”—as I would like to put it—is the external state in the world that consists of his troops marching south: that is, quite literally, his target. His focus is not his own his own refraining from dropping the relevant goal; or his own refraining from adopting inconsistent goals, etc.

Third, and relatedly: intentions, I’ve argued, are realized by a cluster of dispositions. If it were true, as the suggestion goes, that the General intends, for example, to monitor progress towards the goal that the troops march south, then this would imply that the Genral is disposed, among other things, to monitor progress towards the goal of monitoring progress towards the goal that the troops march south. This is simply not plausible. Again, the General’s concern is with the troops marching south. Because that is his concern, he is disposed to monitor progress towards that goal. But his own monitoring of that goal does not itself serve as a goal that he represents, tracks and towards which he monitors progress, filtering out inconsistent ends and figuring out and intending to take appropriate means. Although this is something he is disposed to do, doing it is not his goal; rather, his being thus disposed is in part what it is for him to have the goal that the troops march south.

I think these are compelling reasons to deny that the General intends such things. But the problem for the piggy back theorist is deeper. For, suppose that the General does have the intentions that the piggy back theorist would attribute to him. Presumably, the success of those intentions would amount to the General coming to be disposed not to drop the goal that the troops march south, to track and monitor progress towards it, to not adopt inconsistent goals, etc. The General’s intention that the troops march south, then, could be understood, precisely, as an intention to come to be disposed just as if he intended that the troops march south. To borrow Vermazen’s way of putting this idea, the General could intend to “make himself into, or set himself to be, that kind of input–output device” (1993 p. 238); an input–output device that would mimic an intending device, but directed at the controversial object: that the troops march south.

Well, suppose this is so. The problem then is evident. There is no metaphysical impossibility to the General coming to be disposed as the piggy-back theorist would suggest he intends to be—namely, just as if he intended that the troops march south. That is, there is no metaphysical impossibility to the General adopting such a possible state of affairs as the target state in relation to which he is to guide his thought and action in all the specified ways. But if the General does manage to come to be disposed just as if he intended that the troops march south, then he intends that the troops march south, for this profile of dispositions will play the functional role of an intention, and so realize an intention.

I think this is compelling reason to think that the piggy-back thesis and the Own Action Condition are false. But are there any good reasons to think they are true?¹⁹

Kirk Ludwig (2016, Sect. 7.6) presents an argument in favor of the piggy-back thesis. The argument relies on the idea (famously advocated by Searle 1983) that the content of propositional attitudes that admit of bivalent evaluations is given by their satisfaction conditions. Ludwig calls this the “Satisfaction Principle”. Take beliefs for example. Satisfaction for beliefs—the idea goes—amounts to truth. In other words, a belief is satisfied iff it is true. If so, then the Satisfaction Principle entails that the content of a belief is given by its truth conditions. If a belief is true iff p, then the content of the belief is p. That, in any case, is the idea.

The Satisfaction Principle can and ought to be questioned. It is rejected by different theorists (and rightly so, in my opinion). But it has some initial plausibility, so I shall accept it momentarily for the sake of argument.

The next step is to claim that intentions are also propositional attitudes that admit of bivalent evaluations given by their satisfaction conditions. This seems plausible. Intentions, intuitively, can be satisfied or fail to be satisfied.

Ludwig then claims that the satisfaction conditions of intentions amount to their being successfully “carried out” or “executed”. This sounds plausible, but we should be careful here if we do not want to beg the question at issue. If it is stipulated that “carrying out” or “executing” amounts to the agent doing something that she conceives of as her somehow bringing about the relevant state of affairs, and the content of an intention is given by its satisfaction conditions, then it follows by decree that the content of an intention can only be the agent’s own actions, and that, as Vermazen would put it, apparent “non-act intentions must always ride piggy-back on intentions to act”.

As it turns out, though, this seems to be what Ludwig does in his argument for the piggy-back thesis. Let me cite him at length:

My basic contention is that to intend that q is to intend to bring it about that q… We can employ the Satisfaction Principle to support the analysis. Take the case involving my intending Bud and Pearl to sit next to one another at the party. Under what conditions is this intention satisfied, i.e., executed successfully? Suppose that I form the intention, forget so much as to invite Bud and Pearl to the party, but that they hear about it independently and both crash the party. In the course of it, they find themselves sitting next to one another. Is my intention that they sit next to one another satisfied? My desire that they do so is, but intention is to be distinguished from desire. More pointedly, did I carry out my intention successfully? No, for I didn’t carry it out at all. Why not? It is because I did nothing in pursuit of the end I intended. If I form an intention, for it to be satisfied—carried out—it must be executed, but for it to be executed, I must do something as a result, and not just anything, but something which I see as, or take to be, connected with the end at which the intention aims. (2016 p. 103)

Unlike most theorists’ appeals to the piggy-back thesis or the Own Action Condition as simple items of faith, this is a genuine argument. Nevertheless, it must be noticed that most of the weight is being pulled by the unsubstantiated idea that an intention can only be satisfied if it is “carried out” or “executed”, where what it is to carry out or execute an intention just is for the agent to do something that she conceives of as her somehow bringing about the relevant state of affairs. But why, one ought to ask, should we think that intentions can only be satisfied in this way in the first place? Why—that is—should we think this independently of any previous commitment to the Own Action Condition or the piggy-back thesis? Ludwig does not say.

In a footnote to the cited passage, he says something that could be understood as his reason for thinking this: “Intention—he says—must be distinguished from mere desire. I am urging that the key is that intention always carries with it a commitment to action that constrains further practical reasoning about what to do”. I agree with this remark. What separates intentions from ordinary desires is, indeed, the intimate connection between intention and action. As Bratman puts it: whereas desires are merely “potential influencers” of action, intentions are “conduct controlling” attitudes. But an intention can control an agent’s conduct without leading her to act, and it can do so, just as Ludwig suggests, by constraining her further practical reasoning. To return to our earlier example, the General’s intention that the troops march south can control his behaviour by constraining his practical reasoning so that he does not adopt the further, incompatible goal of cutting provisions. Now, Ludwig mentions that the agent forgets so much as to invite Bud and Pearl to the party. This suggests that the goal does not constrain his thought and action at all. This fact alone would explain the intuition that the intention in this case is not satisfied. We need not appeal to the idea that he does not do anything to bring about the goal. After all, situations could be imagined in which the intention does control his conduct even though it does not lead him to act, if it properly constrains his practical reasoning in all the specified ways. For example, if the agent believes Bud and Pearl will sit together as long as he does not invite Sally, then his intention that they sit together might control his conduct by constraining his reasoning so he does not form the intention to invite her. So we can agree with the spirit of Ludwig’s remark and still be left without a reason to think that an intention can only be satisfied by the agent doing something to bring about the relevant state of affairs.

Now, undoubtedly, there is a way of understanding this phrase in which I think anyone would agree that an intention can only be satisfied by the agent “doing something to bring about” the relevant state of affairs. But although understanding the claim in this sense makes it (as far as I can see) unquestionably true, it takes away most of its substance and translates all the theoretical pressure there is against it towards the Satisfaction Principle—pressure which, I think, the principle cannot then plausibly hold.

To see this, consider the famous case presented by Harry Frankfurt of a driver “whose automobile is coasting downhill in virtue of gravitational forces alone [and who] may be entirely satisfied with its speed and direction [so that] he may never intervene to adjust its movement in any way”. Suppose this driver intends that the car reaches a certain point. Since he is entirely satisfied with its speed and direction, he does not do anything to bring it about that the car reaches the spot. However, throughout the episode, he remains “prepared to intervene if necessary, and [is] in a position to do so more or less effectively”. (1978 p. 160).

Suppose the car does reach the relevant point. Is his intention satisfied? Intuitively, it is. Does he do anything in order to bring this about? Well, he certainly doesn’t do anything that he conceives of as somehow causing or constituting the relevant state of affairs. So, under a strict reading, Ludwig’s idea that an intention can only be satisfied by the agent “doing something to bring about” the relevant state of affairs is false. But, of course, on a much broader understanding, there is something the agent does in order to bring about that goal: he tracks it, monitors progress towards it, filters out inconsistent alternatives as possible ends and remains ready to intervene should he come to believe his intervention is needed. As I will put it from now on: he behaves (where this is broadly understood to include both intentional and non-intentional behaviour, both of thought and of overt bodily action) in a goal-directed manner. So there is no doubt that we could save Ludwig’s principle by understanding it generously enough to make it vacuously true that, whenever an agent behaves in a goal-directed manner towards g, she does something to bring it about that g: namely, she behaves in a goal-directed manner towards g.

I do not think anybody would deny that intentions can only be satisfied by the agent behaving in a goal-directed manner towards the goals they take as objects. So if this is all that is meant by the phrase, I think no one would deny that intentions can only be satisfied by the agent doing something to bring about the relevant end. But the undoubtable truth of this claim is bought at the expense of the plausibility of the Satisfaction Principle, since it is simply not true that, whenever an agent intends a goal, she has her own goal-directed behaviour as a goal. The driver intends that the car reaches a certain spot. His intending this involves his behaving in a goal-directed manner towards that goal. But he need not have his own goal-directed behaviour as a goal. Similarly, the General intends that the troops march south, and he behaves in a goal-directed manner towards this goal. But he doesn’t intend that he behaves in a goal-directed manner towards it. He need not be concerned with his own mind or his own behaviour at all.

If so, then we can’t plausibly hold at the same time that an intention can only be satisfied by the agent doing something to bring about the relevant goal, and that the content of an intention is given by its satisfaction conditions. So I think we can reject Ludwig’s argument.

Having said that, I do think that the Satisfaction Principle should be abandoned. Plausibly, for an intention to be satisfied, not only must its object come about, but the intention itself must play its role in guiding the agent’s thought and action relative to the intended goal, in the manner specified before. Such goal-directed guidance of behavior will sometimes issue in action, if the agent believes such action is required, but it need not, if the agent believes an intervention on her part is not called for.²⁰ It is implausible that such proper functioning is part of the representational content of every intention.²¹

One final reason I can think of for why people might be led to believe that there can only be intentions to do things oneself is the fact that intentions are (or, in any case, are realized by), precisely, dispositions to do things oneself. But that intention is realized by dispositions to do things oneself does not mean that one can only intend to do things oneself. This would be to confuse a fact about the attitude with a fact about the intentional object it can take as content.²²

Intentions are realized by dispositions to think and act oneself in characteristic manners. Dispositions that realize states of mind inhere in individual psychologies, and only directly affect how a single individual thinks and acts. But this fact about the metaphysics of dispositions does not mean that we must see the objects of the agent’s intentions as limited to her own thoughts and actions. Of course, in the most straightforward sense, I can’t be disposed that you do something. My dispositions only directly affect me: what I do, think, feel, etc. But I see no good reason to think that from the fact that I cannot be disposed that you do something it would follow that I cannot intend that you do something, where my intending this would consist in my being disposed to behave in characteristic manners in light of what I intend that you do. We have, after all, the capacity to represent states of affairs that go well beyond our own actions. And we can guide our thought and action in light of those representations (and relevant representations about causal connections). Because of this, an agent can come to be practically committed to a plurality of states of affairs that go beyond her own actions, but her being thus committed will consist in, or be realized by, her dispositions to behave herself in ways that she takes to be conducive to their obtainment.

In the end, I think there are very good reasons to reject the Own Action Condition, and not very good reasons to accept it. So I think we should reject it.

4 Control Condition

Most theorists are happy to allow that we can intend states of affairs that go well beyond our own actions, as long as we believe such states of affairs fall under our control. They endorse some version of what Bratman called the “Control Condition” on intention. This condition, however, is false under any interpretation that would rule out the possibility, to return to the original example, of my intending BEERS.

I should clarify that I do think it plausible that, in order to intend an end e, one must not believe that there is no way in which one can affect the chances whether e. The reason is the following: as I said before, what separates intentions from ordinary desires is the tight connection between intentions and action. Intentions are conduct-controlling attitudes, desires are not. If, however, you believe that there is no way in which you could affect the chances whether e, then it is difficult to see how e could guide or constrain your behaviour as a genuine goal, as a target in relation to which you would adjust your thought and action.²³

Suppose, for example, that you really want it to be sunny today. I presume you believe there is nothing you can or could do that would affect the chances of whether it is sunny or not. If so, then there will be nothing you will or would be disposed to do, or to cease to do, in order to make it more likely that it is sunny. Since you believe that whether it is sunny or not is completely beyond your power of influence, you wouldn’t even remain in a state of readiness to intervene should you come to believe that your intervention is needed. You believe no such intervention could ever be required. This point concerns not only your dispositions to take overt, bodily action. It also concerns your dispositions of thought and deliberation. You do not believe there is or could be anything such that, were you to do it, then because of that it would not be sunny, so there is no possible end you would be disposed to filter out in your practical deliberation. You also do not believe there is or could be anything such that, were you not to intend to do it, then because of that it would not be sunny, so you wouldn’t be disposed to figure out and intend to take appropriate means to it being sunny. This suggests that the state of it being sunny could not be a goal towards which you could come to be disposed in the manner that realizes an intention.

This is why I do think it plausible that, in order to intend an end, you must not believe it is completely beyond your power of influence. I suppose one could say that if someone believes they can affect the chances whether e, then e is somehow under their control. Clearly, though, this is not the kind of control that people who defend this condition have in mind. Otherwise this condition would not rule out the type of ends they want to rule out. I take it, then, that they have something stronger in mind. What exactly would this be?

It could not be that you have complete control over every factor needed for your end to come about. I may intend to go to San Francisco on the train even though I believe I have no control over whether the train comes or not. It seems our intentions can be formed relative to a background of expectations about how the world will evolve independently of our own interventions.²⁴

So the idea is perhaps best put in the following manner: you believe that an end e is under your control (in the sense relevant to the Control Condition) iff you believe (given your expectations about how other relevant things will turn out) that, were you to intend that e, then (at least in part) because of that it would be the case that e.

This makes up for an interesting Control Condition. Among other things, it rules out my intending BEERS, since I do not believe that, were I to intend that BEERS, then (at least in part) because of that it would be the case that BEERS. On the other hand, given my expectations about the reliability of the railroad system, I do believe that, were I to intend to go to San Francisco on the train, then (at least in part) because of that I would go. From now on, I mean “control” in this sense.

So understood, though, the Control Condition is false. To see this, one need just point to intentions that take as content the agent’s own actions, but where the agent does not believe she has such control over them.

Suppose that I intend to defeat you in chess even though I believe you are a much better player. (We could imagine that I intend this because the value I assign to defeating you is so high, relative to the costs I associate with pursuing this end, that the expected utility of adopting and pursuing this goal is greater than that of not doing so, despite my relatively low credence that I will manage to defeat you.) So I do not believe that, were I to intend to defeat you, I would defeat you. This, then, is a case where I do not believe defeating you is, in the relevant sense, under my control. Still, I can intend to defeat you. So one can intend ends one does not believe one can, in this sense, control.

I mentioned before that some people think that it is impossible to intend to do what you do not believe you will do. Once again, though, I can certainly come to be disposed towards defeating you just as if I intended to defeat you—that is, I can be i-disposed to defeat you. This cluster of dispositions realizes the functional role of an intention. So I can intend to defeat you.

If cases like these are possible, then they are enough to show that the Control Condition, so understood, is false. Some philosophers will not accept these cases, because they think it is impossible to intend to do what you do not believe you will do. I cannot say more to convince them. Still, philosophers who believe it is possible to intend to do what you do not believe you will do, may still think that it is impossible to intend recalcitrant social ends. I hope my arguments at least can convince them.²⁵

After all, if it is possible to intend states of affairs that consist of the actions of others, as I argued in the previous section, and it is possible to intend states of affairs that you do not believe you can control, as I have argued in this section, then it is difficult to see why it would be impossible to intend recalcitrant social ends.

In the next section, I will consider, and reject, one last reason to think this.

5 Settle Condition

David Velleman presents a powerful case in favor of the idea that one can only intend what one believes one’s so intending somehow settles. Most importantly, his argument is based on considerations about the metaphysical nature of the attitude in question, and not simply on linguistic proclivities. Intentions—he tells us—are:

attitudes that resolve deliberative questions, thereby settling issues that are up to you. If an issue isn’t up to you, then you are not in a position to settle it, and so you face no deliberative question about it… In resolving this deliberative question, you will arrive at an attitude that settles the issue both actually and notionally. That is, the presence of this attitude will cause the issue to turn out one way rather than another, thus resolving it in fact; while the attitude will also represent the issue as turning out one way rather than another, thereby resolving it in your mind. This issue-resolving attitude is an intention. (Velleman 1997)

I agree with the fundamental idea that drives this remark, but I think that Velleman draws the wrong conclusion. Given the functional characterization with which I am working, intentions are, indeed, “issue-resolving attitudes”, and it is, indeed, true—at least in a sense—that if one is not in a position to settle an issue, then one is not in a position to intend it. The question, however, is what kinds of issues intentions have it as their function to resolve, and I think Velleman goes astray on this point.

There are a few possibilities here. In order to intend to φ, it might be necessary for one to:

• 1. Actually settle (make it the case) that one φs.

• 2. Notionally settle (believe) that one φs.

• 3. Believe one can actually settle that one φs (believe that φing is under one’s control).

• 4. Believe that one can practically settle (resolve the practical question) whether to φ.

Condition 1 is obviously false. I might intend to φ and not succeed in φ-ing.

Condition 2 is false too. Intentions settle, and thereby resolve, practical, or—as Velleman himself puts it—“deliberative questions”, they do not necessarily settle theoretical questions. The question concerning what I will do is a theoretical question. The question concerning what to do is a practical question. Now, in settling a practical question, intentions may also oftentimes help settle or resolve the related, but distinct, theoretical question. But these are still different questions. One can be practically settled on the question whether to φ, and yet be unsettled on the theoretical question whether one will φ.

Condition 3 is just the Control Condition, which I have already discussed and rejected.

Condition 4 is perhaps the most interesting in the present context. One may think that in order to intend to φ, one must see oneself as capable of settling in one’s mind the practical question whether to φ. In other words, one could think that in order to intend to φ one must see oneself as capable of establishing one’s φ-ing as a practical end or goal for oneself. If this were so, then one could likewise think that in order for me to intend that we (you and me) φ, I must see myself as capable of settling, for you as much as for me, the practical question whether we are to φ; that is, settle our φ-ing as a goal for both you and me.

But it is not true that, in order for an agent to intend to φ, she must see herself as capable of settling in her own mind the deliberative question whether to φ. You can come to intend to φ even when you erroneously think you are not capable of coming to be settled on the question whether to φ. You might think you are more ambivalent than you really are on a certain issue. Alternatively, you might not have the conceptual sophistication to even form such a belief in the first place. Plausibly, higher animals and infants are capable of intending to φ even if they lack the conceptual sophistication to form the belief that they are capable of settling the practical question whether to φ. For the same reasons, then, it should also be possible to intend that we φ when one sees oneself as incapable of settling, for everyone involved, the practical question whether we are to φ.

So, if we are going to base the argument for this alleged Settle Condition on the metaphysical nature of intentions, then what this condition, correctly understood, demands, is not that an issue be settled actually, nor that it be settled notionally, nor that one believe one can settle an issue actually, nor that one believe one can settle an issue practically. What the condition would demand is simply that, if one intends to φ, then one is settled on the practical question whether to φ. Formulated in this way, this condition is obviously true. It just describes what it is to intend. To intend is to be settled on a practical issue. If one cannot be practically settled on an issue, then one cannot intend it. But this doesn’t rule out my intending that we get together for beers on Friday, since our getting together for beers on Friday can be, for me, a settled object of pursuit.

6 Concluding Remarks

I have argued that one can intend recalcitrant social ends. Regardless of the specifics of these arguments, I hope that the central idea that drives my thesis is clear: we have the capacity to represent possible states of affairs that go well beyond what we believe we can do or control, and we have the capacity to set those possible states of affairs as targets in relation to which we adjust our thought and action in precisely those ways that philosophers of action have told us are characteristic of intentions, and that would play the functional role of intentions.

We can, no doubt, reserve the word “intention” to refer to cases in which we are so disposed towards ends that are neither social nor recalcitrant (or perhaps just not both social and recalcitrant). We could refer to the state of being so disposed towards social or recalcitrant ends simply as the “state of being practically committed to an end or goal”. This would be perfectly fine as far as I am concerned, as long as we recognize that the underlying metaphysical structure (and the associated rational dimension, although this was not my focus in this paper) is (are) the same in both cases. My concern, then, is not with the word “intention”, but with the psychological structure (and the associated rational dimension) that theorists of action are pointing to when they use that word. And what is important to emphasize is that that structure is common to both kinds of states.

Put differently: there is an interesting leap, both metaphysically and rationally speaking, between the mental state we ordinarily call a “desire”, and the mental state we ordinarily call an “intention”. There is no equivalently interesting leap, either metaphysically or rationally speaking, between the state we ordinarily call an “intention”, and the state I am now referring to as a “practical commitment” to a social or recalcitrant end. This state is just like an intention except for the fact that it takes as contents social and/or recalcitrant ends. As long as this is clear, I wouldn’t have any objection to the linguistic stipulation that one can only intend one’s own actions, or what one believes one can settle or control.